15. Video QoS Analysis over Wi-Fi Networks

According to the recently published report (June 1, 2011) “Cisco Visual Networking Index: Forecast and Methodology, 2010–2015,” global Internet video amounts to over 40% of the consumer Internet traffic, predicted to reach 50% mark by the end of 2012. The use of advanced video communications will increase in the enterprise sector causing business IP traffic to grow by 2.7 times between 2010 and 2015 (the forecast period). Business video conferencing will increase six times over the forecast period. The collective video traffic, such as TV, video on demand, and P2P, will reach approximately 90% of global consumer traffic by 2015. On the wireless and mobile side, Internet traffic originating with non-PC devices will grow five times over the forecast period; and by 2015, traffic from wireless devices will exceed traffic from wired devices. Globally, mobile data traffic will increase 26 times over the forecast period. Furthermore, machine-to-machine traffic will grow at 258%.

These trends substantiate the view that a range of video-based multimedia services^* will be delivered over multiple heterogeneous network platforms, comprising fixed, wireless, and ad hoc networks,^† all employing IP-based technologies. In the recent past, it was not feasible to deploy many of these emerging multimedia services due to the lack of business models and the huge investments that were required. The rising demands for mobility and anytime, anywhere communications (e.g., military battlefield operations, emergency, and disaster applications) and the increasing amount of mobile traffic and devices have transpired and accelerated the developments of ad hoc networks allowing shorter network deployment times at lower costs. Therefore, future communication systems will increasingly integrate and rely on ad hoc networks in order to support anytime, anywhere seamless mobility.

While multimedia forms high data rate traffic with stringent quality of service (QoS) requirements, wireless ad hoc networks characterize frequent topology changes, unreliable wireless channel, network congestion, and resource contention. We are motivated by the many challenges and opportunities in supporting multimedia over heterogeneous networks with scalable QoS and have developed a range of design and analysis techniques for wireless infrastructure and ad hoc networks in order to support multimedia applications with scalable QoS.

15.1.1 Objectives

The aim of this chapter is to present an overview of our work on multimedia wireless networks. Firstly, an extensive review of the literature on multimedia ad hoc network design and QoS analysis is presented. Based on the review, we structure the literature on multimedia ad hoc networks design and QoS into 10 dimensions, including routing algorithms, QoS, medium access control (MAC) design, scalability, security, green networking, and so on. The literature on each dimension is discussed in some detail. Secondly, a detailed analysis of multimedia applications over wireless networks, both infrastructure and ad hoc, is presented. For ad hoc networks multimedia QoS analysis, four routing schemes are used: ad hoc on demand distance vector (AODV); optimized link state routing (OLSR); hierarchical clustering, provisioning, and routing (HCPR); and geographic routing protocol (GRP). Several networking scenarios have been carefully configured with variations in network sizes, applications, codecs, and routing protocols to extensively analyze the ad hoc networks performance. All the network analyses presented in this chapter are based on simulations using the OPNET simulator. The HCPR scheme is implemented as an extension (module) to the OPNET simulation software.

We note that the vast majority of networks are populated with multimedia traffic, as opposed to video alone, and therefore we have included voice and data (HTTP traffic) in our network analyses. Nevertheless, we will present network analysis with focus on video application alone. Our approach is to simulate and evaluate realistic multimedia application scenarios for networks with a sufficiently large number and type of applications, both elastic and nonelastic.

The rest of the chapter is organized as follows. Section 15.2 briefly provides the necessary background material related to this chapter. Section 15.3 provides the literature review on multimedia ad hoc networks design and QoS structured into 10 dimensions. Section 15.4 presents the multimedia QoS analysis over wireless networks. Finally, Section 15.5 concludes this chapter with some directions for future work.

15.2 Wireless Networking Technology, QoS, and Routing

This section provides the background material related to this book chapter. We briefly introduce IEEE 802.11 local area wireless network (WLAN), the wireless networking technology that we have used in our network analysis work. Subsequently, we introduce QoS, its measuring metrics, and routing schemes used in this chapter.

The IEEE 802.11 WLAN, also known as Wi-Fi, is widely used these days because of its flexibility and the low cost of equipment and deployment. The IEEE 802.11 is a family of standards that specifies the MAC and Physical layers of a WLAN (O’Hara and Petrick, 2005). The main purpose of 802.11 standard, according to the published standard (IEEE 802.11 Working Group, 2007), is to provide wireless connectivity for mobile and stationary devices within a certain range. WLAN can be configured in two modes: infrastructure mode or ad hoc mode. In this chapter, the WLAN performance in both modes is analyzed in detail.

15.2.1 QoS Metrics

QoS is the performance level of a service offered by the network to the application (Murthy and Manoj, 2004). Applications may vary in their requirements from the network; some of them need to have fast packet delivery but could be flexible in security or confidentiality of the packets. The QoS is the service that prioritizes users’ requirements and then aims to guarantee satisfying their requirements without effecting other users’ requirements. Some common QoS parameters for applications include delay, jitter (variation in delay), throughput, and (tolerance in) packet losses. Networks create delay, jitter, and packet losses due to, for example, limited bandwidth availability, buffering and switching delays, lack of buffer space, and transmission errors. An application can use its QoS parameters to negotiate QoS with the underlying network. The network designer or operator can consider these QoS parameters for network design purposes to negotiate service level agreements or to develop and implement policies and procedures required to guarantee service level agreements. This has led to many developments and standardization activities to provide end-to-end QoS over the Internet, including the IPv4 to IPv6 evolution. Best effort networks are not enough for some applications where, for example, delay cannot be tolerated, and therefore there have been many methods and techniques used to overcome such problems. Each QoS-enabling method has its own pros and cons and there is no technique that can satisfy all QoS requirements. Well-known QoS techniques include buffering, shaping, resource reservation, admission control, multipath routing, packet scheduling, and provisioning. QoS architectures developed to support end-to-end QoS include integrated services (Braden et al., 1994), differentiated services (Blake et al., 1998), and multi-protocol label switching (Rosen et al., 2001).

A number of QoS metrics exist to measure network performance. These are described in the following. We mainly focus on those that we have used in the analysis of results presented in this chapter. Traffic sent is the traffic that has been successfully sent from the application layer to the next lower layer (transport layer). Traffic intended to send is the traffic that has been planned to send from application layer to lower layer. However, not all traffic is intended to send is sent because some layers (e.g., physical) will not be able to handle the traffic for any reason. Throughput is the percentage of the average traffic received per second to the traffic that was intended to send per second in the network. In results presented in this section; throughput is used for every individual application: video, voice, and HTTP, besides having throughout of total traffic in network. Delivery ratio (DR) is the percentage of the average traffic received per second to the traffic sent per second in the network. In this section, there will be DR for every individual application: video, voice, and HTTP, besides having the total DR of all traffic in network. End-to-end delay is the average time spent in seconds for the packet to reach destination. The video and voice end-to-end delay have slightly different definitions by OPNET. The video end-to-end delay is defined as the average time spent in seconds to transfer video packets from source’s application layer to destination’s application layer in the network. It is measured from the time of creating the packet until the time of receiving it. The voice end-to-end delay is the average time in second that voice packet takes to reach destination from time of encoding the analogue signals at source to the time the packet is decoded at destination. This is called mouth-to-mouth delay, which include encoding delay, compression delay, network delay, decompression delay, and decoding delay (OPNET Technologies Inc, 2008). Delay variation is the average variance in packet end-to-end delay. In OPNET, user can collect delay variation for video and voice packets besides having jitter for voice. Voice jitter is the variation of delay in second in received voice packets (Cisco Systems, 2006). If two packets left the source voice application at t_1 and t_2 times consequently and arrived at destination’s voice application at t_3 and t_4 consequently, the jitter is the result of this following equation: (t_4 – t_3) – (t_2 – t_1) (OPNET Technologies Inc, 2008). Voice mean opinion score (MOS) “is a subjective measurement representing the quality of digital multimedia including video, voice, or audio” (Mehmood et al., 2011) and it represents the quality in a numerical format ranging from 1 as reference to bad quality to 5 as reference to excellent quality. In the results presented in this chapter, the average of voice MOS in the network per second was taken and used. HTTP page response time is the average time in second required to retrieve the entire page with all objects (e.g., image) in the page.

15.2.2 Routing Protocols

In data networking, the process of identifying and selecting network routes to direct the network traffic is known as routing. A large number (over 30, see Liu and Kaiser [2003], for instance) of routing protocols exist which mainly differ in the way they select the network path; however, most of them fall into two major protocol categories: Distance vector protocol and link state protocol. Moreover, routing protocols can also be classified on the basis of how ready they are to send the packets into reactive and proactive protocols. There are also some other classification methods in the literature, such as whether the routing protocol is flat or hierarchical, source routing or hop-by-hop, location based or not, uniform or nonuniform (Liu and Kaiser, 2003), and single or multiple channels.

In this chapter, we have used four routing schemes to analyze and compare video performance over wireless networks. These are the AODV (Perkins et al., 2003); the OLSR (Clausen and Jacquet, 2003); the GRP (Takagi and Kleinrock, 1984); and the HCPR scheme (Alturki and Mehmood, 2012; Mehmood and Alturki, 2011). The AODV protocol was first published as an Internet draft by Charles Perkins in 1997, then in 2003 published as an RFC (Perkins et al., 2003). It is a reactive routing protocol that is based on a distance vector routing approach. Nodes are kept quiet until a connection is required. The node that requires the connection will broadcast a route request message asking for a route to destination. A reply to this message is expected from any node that has the route to destination or from the destination itself. Subsequently, it will choose the route to destination with the least number of hops. The OLSR protocol (Clausen and Jacquet, 2003) is a link state proactive routing protocol that maintains its routing and forwarding table with disregard to packet arrivals. It sends a periodic hello message to exchange information about network. In order to avoid extensive routing maintenance messages from consuming the limited wireless bandwidth, it has some chosen nodes called Multi point relays (MPRs) to reduce packets. Those MPR nodes are selected nodes that do two main jobs: (1) generate topology control messages and (2) act as data packet forwarders for other nodes. The advantage of using MPRs is clearly observed when a network is big and dense in comparison to pure link state routing or other ad hoc networks routing protocols. Also, when a network is small and nodes are sparse, OLSR becomes a pure link state routing protocol. Packets are routed in a hop-by-hop basis—that is, nodes in the route will make the forwarding decision according the local routing table. GRP is a class of protocols that use geographical position for routing packets, firstly proposed in Takagi and Kleinrock (1984). We have used the implementation of the GRP protocols as available in OPNET.

The HCPR was proposed in Alturki (2011), Alturki and Mehmood (2012), and Mehmood and Alturki (2011). HCPR is a cross-layer scheme that works on application, transport, and network (routing) layers. The HCPR was designed with particular emphasis on scalability and in order to address QoS challenges to deliver multimedia over ad hoc networks. The HCPR scheme is based on intelligent protocols that optimize multimedia QoS provisioning by enabling and exploiting interactions between application, transport, and network layers. Most of the cross-layer schemes for multimedia ad hoc networks have focused on interfacing or merging, the physical and MAC layers; MAC, physical, and network layers; MAC, application, and physical layers, etc. We believe that no other work has identified the potential of, and have considered, the delicate balance of layered interaction between application, transport, and network layers, as is the case in HCPR. The interfacing of application, transport, and network layers in HCPR allows HCPR to scale well, and provide QoS, to large networks under heavy multimedia traffic. The HCPR scheme comprises two major phases: the network formation phase and the network operational phase. The network formation phase prepares the network in order for the network to move to the operational phase. The network formation phase enables the nodes to find each other, their positions, to form node clusters and an overlay network connecting the cluster heads (i.e., the head or leader of a cluster). The network formation phase is divided into the network discovery phase and the cluster overlay formation phase. The network discovery phase allows the nodes to know the number of nodes, their respective positions, and the geographic boundaries of the network. The cluster overlay formation phase is used to build multiple network clusters, elect cluster heads, and build a QoS routing overlay network. Subsequent to the formation of the HCPR-enabled network, the network moves to its operational phase: a node in the network can now make requests to connect to a destination node with some application. A node wishing to receive a service, however, does not talk to the destination node directly; rather it requests the cluster head of its cluster to make a request to the destination on its behalf. This request is initiated from the application layer, which will call the corresponding function of HCPR (i.e., at the transport layer) that makes the reservation request to the cluster head. Essentially, in an HCPR-enabled network although the requests are made locally, these are propagated and provisioned by a hierarchical structure made up of cluster head nodes. Consequently, the network provides high reliability and scalability. In a mobile ad hoc networking environment, the network will continue to toggle between the network formation and the network operational phases because it will have to reconfigure itself according to the changing network conditions. Further details of the HCPR scheme, its protocols, and its implementation as a separate module of the OPNET simulator can be found in Alturki (2011) and Alturki and Mehmood (2012).

15.3 Multimedia Ad Hoc Network Design and QoS—10 Dimensions

We now provide an extensive review of the literature on the main topic of this chapter, that is, multimedia and ad hoc network design and QoS. A huge amount of literature is available on the design and performance analysis of multimedia ad hoc networks. We structure the literature on ad hoc networks design into 10 dimensions. The 10 dimensions are routing algorithms, QoS, MAC design, scalability, cross-layer design, security, green networking, topology control, wireless sensor networks (WSNs), and methodologies for multimedia performance analysis over ad hoc networks. We shall define and discuss each dimension in a separate section and, wherever available, we shall include a review of the literature on each dimension in relation to video or multimedia QoS support and analysis.

15.3.1 Cross-Layer Design in Multimedia Ad Hoc Networking

The traditional layered protocol network architecture is based on a series or stack of layers, each layer provides services to the higher layers, allowing an abstraction, and hiding implementation complexities, for the higher layer. This layered approach reduces complexity through modularization of the network and allows manageable design, interoperability, extensibility, scalability, and standardization. There are, however, well-known disadvantages of the layered design. The so-called cross-layer design techniques have emerged in the recent years as a potential solution to address the multimedia QoS provision challenges plaguing the development of heterogeneous networks. The cross-layer ideology is to optimize the design and performance by increased and effective interaction between layers and optimization at a global level (multiple layers) rather than at the local level (single layer).

Ramanathan and Redi (2002) have described the cross-layer approach as a promising method that could solve problems related to QoS. By allowing layers to use other layers’ information and accessing them freely, layers could adapt to the changes using other layers’ information. Therefore, the QoS requirements can be satisfied accordingly. Lee and Song (2010) have proposed a cross-layer algorithm for video streaming over ad hoc network. The algorithm chooses the efficient PHY mode and the retransmission limit of WLAN for each node in a distributed way. The algorithm makes the choice depending on the information available at other layers, application, MAC, and PHY. Pompili and Akyildiz (2010) have studied the multimedia underwater delivery in the acoustic sensor networks. They mentioned that the previous underwater communication works have followed a traditional stack layered approach, which was firstly designed for wired networks. Authors claim that the cross-layer approach will improve the multimedia and delay sensitive applications in underwater communication environments. They studied the interactions of underwater communication functionalities like modulation, forward error correction, MAC, and routing. In addition, they proposed a new distributed cross-layer solution that allows devices to share the underwater acoustic medium fairly and efficiently which is characterized by having high delay and bandwidth limitations. Gharavi and Ban (2003) have proposed a cross-layer feedback control scheme for video transmission over ad hoc networks. The proposed scheme was designed to overcome the current problems of variations of delay and bandwidth. The current schemes like real-time transport protocol have limitations in mobile environments which causes the network to be congested. The proposed scheme is zero-delay and traffic free channel assessment scheme that allow the sending node to adapt to the current network conditions. The scheme takes the critical information (such as number of hops) from the underlying routing layer. Those extracted information are also used in other proposed schemes like packet recovery and redundant packet transmission.

Melodia and Akyildiz (2010) have proposed new cross-layer communication architecture for multimedia delivery over WSN based on time hopping impulse radio ultra wide band. The architecture aims to deliver heterogeneous application traffic over WSN reliably and with flexibility by controlling the interactions between the layers. The resulted simulations show that the architecture achieved the aimed target without affecting the modularity of the stacked design. Setton et al. (2005) have investigated the potential that cross-layer approach can offer to support real video by allowing layers to exchange information and showed how end-to-end performance was optimized by adapting to such information. The results show that the video-stream performance has improved substantially. Mundarath et al. (2009) have proposed a new cross-layer approach for multiple antenna ad hoc networks that consider QoS and reduce power consumption called QoS-aware smart antenna protocol. The proposed approach adapts the degree of freedom present in the multiantenna in each node to achieve less energy consumption while keeping application’s QoS requirements assured. The results show that using cross-layer approach in the design saves considerable energy consumptions in comparison to other QoS schemes that use strict layering approach.

15.3.2 Routing Algorithms

Routing in ad hoc networks has received significant attention from the research community, most probably because ad hoc networks’ main concept is based on routing, where nodes are responsible for routing the packets for other nodes. Despite extensive work in the field of routing in ad hoc networks, routing is still a fundamental problem in ad hoc networks’ design. The first problem is the scalability of routing. Most current routing protocols have only been tested for relatively small size networks, yet none of the current routing protocols have showed reasonable performance with large networks and heavy network traffic. Another problem in designing routing algorithms for ad hoc networks is the large number of design parameters that are supposed to be optimized such as mobility of nodes, range of applications and their intensity, low power consumption, etc. We now review some of the most notable literature on multimedia ad hoc networking where the focus of the researchers has been to either design/improve routing protocols or study routing performance against multimedia traffic.

Considerable amount of research in ad hoc networks has been done to compare and analyze various routing protocols. Layuan et al. (2007) have carried out a multimedia performance analysis of four routing protocols using QoS metrics, routing load, and connectivity. QoS metrics include delay, jitter, throughput, and loss ratio. They have used different network sizes of 10, 20, 40, 50, finishing with 100 nodes, and placed randomly in 1 × 1 km area. They have used NS2 as a simulation tool and used user datagram protocol (UDP) constant bit rate (CBR) traffic with 512 byte as the packet size and pairs of 6, 12, 24, 30, and 60 UDP streams. Ng and Liew (2007) have studied and indentified the maximum throughput in 802.11 multi-hop networks. They mention that having the network overloaded leads to two problems: packet loss and rerouting instability. They used NS2 to show their results, which is based on having eight stationary nodes placed in string or chain topology. The transmission range was 250 m and UDP stream traffic was sent from the first node on the edge to the last node on the other edge with 1460 byte packet size. The main metric of evaluation was throughput in percentage. Their results show that 1.18 Mbps is the optimal sending rate. Kumar et al. (2009) have compared two reactive routing protocols, DSR and AODV, by varying the traffic load, mobility, and type of traffic using NS2 simulator. They used packet delivery, packet loss, and end-to-end delay as metrics to measure the performance. They had generated four traffic patterns with randomly varying the sources and destinations using two type of traffics CBR and TCP. They used 50 nodes placed in a 1.5 × 0.3 km area. The packet size was 512 byte. A random waypoint mobility model was used during those emulations.

Pucha et al. (2007) have studied the performance of ad hoc networks under different traffic patterns. The authors argued that previous performance analysis studies cannot give the same results with different traffic patterns. For this reason, the authors studied the impact of traffic patterns on the performance of 112 nodes. They used their new connection models to allow multiple connections per source. They studied three routing protocols: DSDV, DSR, and AODV using NS2 simulator, where 112 nodes were populated in 2250 × 450 m area with 250 m transmission range and 2 Mbps bandwidth. Around 80% or 90% of nodes were CBR traffic sources of 60 packets/s and the packet size chosen was 64 byte. They used packet DR and delay as valuation metrics. Qadri and Liotta (2010) surveyed the performance analysis studies of ad hoc network routing. The study gives a good background of ad hoc routing classification and categories and descriptions of some common routing protocols. This is followed by an overview of ad hoc network requirements and metrics used for performance analysis. We note here that all papers and work discussed in this chapter (Qadri and Liotta, 2010), which was published fairly recently (in 2009) were general in their scenario settings and analysis and there were no detailed specifications in the scenarios for video, voice, and data together such as the one presented by us. Chatzistavros and Stamatelos (2009) have examined the behavior of ad hoc network routing protocols through the simulation using CBR traffic with different packet sizes and different topologies. They have used QualNet to compare DBF, DSR, and ZRP routing protocols and to study the mobility effect on performance using 802.11 standard. They started by simulating a chain of nodes and one flow of CBR traffic between the two ends and then made parallel chains of nodes with a CBR flow with every chain.

15.3.3 Multicasting and Quality of Service

Satisfying QoS in wireless infrastructure networks still has some challenges to meet (Murthy and Manoj, 2004) due to, in particular, the unpredictable nature of the wireless medium. Because ad hoc networks use wireless technology, it inherits this open problem. What makes achieving QoS requirements in ad hoc networks harder is its lack of infrastructure and its multi-hop functionality. It is proposed in literature that to improve the QoS in ad hoc networks, the function of QoS should be taken from the upper layers to be placed in the MAC layer (Ramanathan and Redi, 2002). Also, some others researchers propose that QoS cannot be achieved in the traditional structured layering approach but, instead, by collaborations between the layers by allowing layers to interact with each other and use other layers’ information.

de Morais Cordeiro et al. (2003) presented a comparison of well-known multicasting algorithms in ad hoc networks. Morgan and Kunz (2005) proposed a QoS gateway that acts as an interface between the two types of networks. Their motivation was that ad hoc networks sometimes need to connect to some public service on the Internet which probably requires guaranteeing the QoS requirements. Their proposed work is based on having an existing QoS solution in both network types, such as SWAN/ESWAN in Ad hoc networks and DiffServ in structured networks. The main job of the gateway is to map the QoS in the ad hoc network to the QoS in the infrastructure network. The proposed gateway has shown smoother bandwidth when the gateway is used between the two different networks in comparison to not having any QoS gateway. A description of the available approaches for group communication in MANET, like multicasting and broadcasting, is covered in Mohapatra et al. (2004). It has been mentioned that QoS in group communication is an open problem. The standard QoS protocols are required to guarantee some measurable performance, including delay, bandwidth, packet loss, and delay variance. MANETs add two more attributes—power consumption and service coverage. Those attributes and some characteristics of MANET make the QoS in MANET a complex process to be achieved. Not only is QoS desirable for MANET to guarantee general aspects of its communication but also group communication needs QoS support that can be customized to suit and satisfy the group communication protocols. One of the proposed QoS protocols is QoS-aware core migration for the multicasting algorithms. It uses a group of shared multicast tree in which the leaves achieve the required multicast quality.

15.3.4 Medium Access Control Design

MAC layer has two main functions: addressing and access control. WLAN 802.11 provides MAC functionality for wireless local area networks. The exiting MAC protocols, like 802.11, do not very well suit ad hoc networks because the existing protocols have been firstly developed for specific types of networks; and all of them are infrastructure networks. Designing a MAC layer for ad hoc networks requires the consideration of many issues particularly related to its lack of infrastructure. Some of these issues have been discussed separately in this literature review as separate problems, such as security, QoS, and power consumption. Other issues are related to MAC design only and these include collision control, channel optimization, and channel fairness. The integration of the MAC sub-layer with the other network layers in MANET has been discussed in the literature. This area of literature falls under cross-layer protocols and has been discussed in Section 15.3.1.

15.3.5 Scalability

Scalability in ad hoc networks is an extremely important and active research area. The network scalability can be defined in terms of various network parameters such as the number of nodes in the network, the number of multimedia flows, and their intensities, etc. Scalability in ad hoc networks design is difficult due to its infrastructure-less attributes (e.g., nodes acting as routers, multi-hop), and hence we can say, in general, that the network performance has an inverse relationship with the number of nodes in the network. Researchers in the ad hoc networks area are trying to answer the question: to what extent can ad hoc networks grow and how can that be extended while maintaining an acceptable level of performance?

15.3.6 Security in Ad Hoc Networks

Security has emerged as one of the most important topics in every aspect of computing and communications, from operating systems to networking and programming languages. The level of security required differs from one application to another. For example, the level of security required for a gaming PC is much smaller than the requirement for securing internet banking web server. Similarly, in ad hoc networks, applications differ in their security requirements, for instance, securing an ad hoc network system for military use is not comparable to securing an ad hoc network for cooperative gaming. The main security goals as described in Lidong and Haas (1999) are: availability, confidentiality, integrity, authentication, and non-repudiation. There are many security methods in practice to achieve security goals. These methods include, among others, authentication protocols, digital signatures, and encryptions. Although the traditional security technologies do play a positive role in achieving many of the security goals; however, they are not sufficient to achieve the desired highest levels of security in ad hoc networks. Two other methods used in ad hoc networks to provide security are route redundancy and distribution of trust over the nodes. Route redundancy promotes the availability of the network connection. If one route has fallen down, packets are able to go to the destination using another route. This feature makes the network connection more available. The distribution of trust over the network nodes promotes availability and the authentication. Because of the low physical security and availability of the connection, nodes in MANET are not trustworthy. Consequently, trust can be distributed over the network nodes and the aggregation of the nodes’ trust will be trustworthy.

15.3.7 Energy and Computational Efficiency: Green Networking

Until recently, there has been a lack of literature on energy and computational efficiency in ad hoc networks design. However, due to the increasing emphasis on environmental issues and “green” approaches to everything, analysis and efficiency of energy usage is becoming a critical topic. Video and multimedia uses huge bandwidth and, to date, there is virtually no effective work available on minimizing energy usage in multimedia ad hoc networking. However, research in this area is expected to grow rapidly in the near future.

15.3.8 Topology Control

This challenge is a direct result of having mobile nodes in the network although topology can also be affected in a static network due to nodes becoming faulty or unavailable. The challenge is to design a network that maintains its performance as if there is no change in the network topology. This challenge could be achieved by predicting the movement of nodes and their neighbors. As a result, many researchers are investigating modeling of nodes’ mobility while others are focusing on the methods and techniques used to control the mobility of the nodes to give an acceptable performance. Camp et al. (2002) have given comprehensive coverage of mobility models in ad hoc networks. Broch et al. (1998) in their work in comparing the performance of four well-known routing protocols have shown how these protocols are affected directly by the movement pattern of the nodes. When nodes are highly mobile, protocols showed the worst performance, and when nodes have less movement, the performance increases dramatically. Hong et al. (1999) have surveyed the mobility models in cellular and ad hoc networks. They have proposed a group mobility model called reference point group mobility. Regarding a cellular network, the authors have mentioned Random Walk and Random Gauss–Markov models which have been used in some people’s work to simulate node movements. In ad hoc networks, Random Walk, Random Waypoint, Chiang’s Markovian, Pursue and Column models have all been used in simulating the individual nodes.

15.3.9 Wireless Sensor Network

WSNs are a type of ad hoc networks, which inherit the ad hoc challenges in addition to some challenges of its own. WSN is an ad hoc network that has the ability to sense some environmental variables. WSN nodes are used to collect data about environmental variables and then inform a sink node or other nodes about their sensed data. WSNs are growing in their applications and scope, their most recent applications relevant to the topic of this chapter (i.e., multimedia over ad hoc networks) is the use of mobile devices as sensor networks in emergency situations such as in natural or manmade disasters, data search and analysis, etc. (Alazawi et al., 2011; Satyanarayanan, 2010).

15.3.10 Multimedia Ad Hoc Networks: Analysis Methodologies

This section presents the various works from the literature that present an analysis of multimedia performance over ad hoc networks by studying the existing networks and protocols or proposing new ones. We begin with the various approaches that attempt to improve multimedia performance over ad hoc networks by new proposals.

Literatures that have proposed new methods to improve multimedia over ad hoc networks and evaluate it include Hongqi et al. (2008), Li et al. (2000), Li and Cuthbert (2005), Utsu et al. (2010), and Xue and Ganz (2003). In Hongqi et al. (2008), the authors have proposed a new method to reduce packet loss, whereas the authors in Utsu et al. (2010) propose a new approach to make video transmission smoother. Li and Cuthbert (2005) claim that they have made improvements to the ad hoc performance for multimedia. Xue and Ganz (2003) propose a new end-to-end QoS protocol, while the authors of Li et al. (2000) propose a scalable location service. Each of these works will now be discussed in more detail. Hongqi et al. (2008) have studied the performance of voice over wireless ad hoc networks and proposed a new scheme to improve the performance by reducing packet loss. They have used MOS as the main measure for the performance, besides packet loss and jitter. Utsu et al. (2010) have studied the problem of video over ad hoc networks and suggest new methods to make video transmission smoother. In order to evaluate their results, they used NS2 as a simulation tool with some experimental work. The number of nodes simulated in NS2 scenarios was 20 nodes placed in 1 × 0.6 km and a random waypoint mobility model was used. The bandwidth used was 2 Mbps with a 250 m transmission range. They used DSR as routing protocol and a video traffic application of 128 kbps and frame size of 1024 byte, and bit sequence traffic to simulate that. Throughput was the main evaluating metric. Li and Cuthbert (2005) have proposed a new QoS multipath routing method that improves the performance of multimedia. They have shown some performance improvements by simulating 50 nodes in 1 × 1 km area and 250 m transmission range using OPNET 8.1 modeler. Traffic was populated with 512 byte CBR data packets. Five to 20 sources were chosen randomly to send 80 kbps expedited forwarding packets. Another 20 nodes sent 8 kbps background packets in best effort manner. The source-destination pairs were chosen randomly with variety in the number of connections for EF packets. They used packet DR and packet end-to-end delay as metrics for performance evaluation. The simulation was run for 800 s.

Li et al. (2000) propose a distributed location service for mobile ad hoc network routing that scales well when the network gets bigger. They use a NS2 simulator with up to 600 nodes and 2 Mbps bandwidth with a 200 m transmission range. The traffic is CBR with number of connections half the number of nodes. For each connection, four 128-KB data packets per second are sent to a destination for 20 s. Connections are generated at random times in the simulation time. Packet DR metric is used to view the traffic performance. Xue and Ganz (2003) have introduced a new resource reservation–based routing and signaling protocol called ad hoc QoS on-demand routing, which provides end-to-end QoS support and has been implemented in OPNET modeler. They have used 802.11 DCF-enabled nodes with 2 Mbps and 250 m transmission range, and they have used small and large network sizes to see how the protocol behaves. Small networks are used to discover the QoS recovery time which consists of five nodes—four statics and one mobile with one flow (400 kbps) from source to destination. The large network is used to study the performance of the proposed protocol and consists of 50 nodes placed in a 1 × 0.5 km area and moving randomly. To simulate multimedia streaming, they have used CBR traffic with 512 packet size and sent 10 packets a second. They have populated 10 and 15 traffic flows with random sources and destinations. In order to study performance, they used the following metrics: traffic admission ratio, end-to-end DR, average end-to-end delay, ratio of late packet, and normalized control overhead. They defined end-to-end DR as the “ratio between the number of data packets received at the destinations and the number of data packets sent from the sources. This metric indicates the reliability of the admitted flows.” The average end-to-end delay defined as “the latency incurred by the packets between their generation time and their arrival time at the destination. This metric indicates the performance of the admitted flows.”

We now turn to the second category of works that focus on the analysis of multimedia performance over wireless and ad hoc networks. Sondi et al. (2010) have done a performance evaluation of OLSR-QoS extension by using OPNET simulator. They populated 50 nodes in a 1 × 1 km area and nodes were equipped with an 802.11 g network card. They have simulated scenarios with statics and mobile nodes, with the aim of evaluating voice communication with some background. Therefore, they have made one voice connection between two nodes using G.711 codec (64 kbps) with the two nodes having FTP traffic as background traffic. For evaluation, they have used IP number of hops, jitter, packet end-to-end delay, MOS, and traffic sent and received as evaluation metrics. It has been observed that voice packet end-to-end delay were around 0.06 s in most scenarios. Gottron et al. (2009) show the feasibility of voice communication in larger scale ad hoc networks if the right settings are used. They have used Jist/SWANS simulation tools to conduct their studies and scenarios with 50,100,200, and 500 nodes. They have evaluated the results according to (1) Voice quality where MOS is used as the measure, (2) Transmission delay, and (3) Packet loss. They have varied the network load between 5 and 20 voice streams of G.711 voice codec. The transmission range used was 250 m and nodes were surrounded by an average of eight nodes. There is no clear description of how the source and destination were chosen.

Jeong et al. (2009) have studied the observation that only five simultaneous VoIP calls can be made in 802.11 using NS2 and show the reason behind that. Then, they proposed a new algorithm at the MAC sub-layer to improve performance. They have used jitter, loss rate, delay, and call capacity as metrics for measurement. They used a network consisting of wireless and wired nodes. Two laptops communicated a single VoIP application over the wireless network and the rest of the communications were within the wired network. They report that theoretically 85 calls can be made in 802.11b but, in practice, only five can be supported with QoS.

Hofmann et al. (2007) have carried out a performance analysis study of voice over a static wireless ad hoc network. They used throughput, delay, and jitter as metrics for the evaluation using simulation (NS2) and experimental work. Eight nodes have been placed to form a ring topology and a source is sent to the destination node which is the last node in the ring and should be accessible through the other six nodes. They used 1 Mbps and G.711 codec. Santos (2009) has studied the performance of VoIP over ad hoc networks using OLSR routing protocol. It varied node intensities, number of data streams, and mobility in the simulations to obtain 18 different scenarios using OPNET simulator. It focuses on studying the impact of the number of nodes, number of streams, and node mobility on the performance of ad hoc networks using end-to-end delay and packet loss. The author has simulated 10–50 nodes placed randomly in a 1 × 1 km area to send streams of VoIP with fixed packet lengths of 200 byte using the G.711 codec. The author has defined the end-to-end delay but not packet loss. However, the author did not show clearly how end-to-end delay was obtained and whether voice application or IP statistics were used. Their definition of end-to-end delay is defined as delay measured from the instant a packet leaves the sender’s Network Interface Card (NIC) to the instant it is received at the destination’s NIC. Still, it is not clear how that has been obtained from the OPNET simulator, and from our experience, there is nothing defined as such in voice application.

An extensive literature review that we have carried out have confirmed our view that the existing methodologies to evaluate multimedia networking studies lack sufficient details and fail to provide assessment of realistic multimedia networking environments. To address this, we have developed a novel analysis methodology for multimedia networks (the methodology is used in our research, see Alturki and Mehmood, 2008; Mehmood and Alturki, 2011; Mehmood et al., 2011; and this chapter for the QoS analysis presented in Section 15.4). The approach is to simulate and evaluate realistic multimedia applications scenarios for networks with a sufficiently large number and type of applications, both elastic and nonelastic. The network QoS is explored for a set of different applications (as may typically be found in realistic situations), so that the mutual effects of network applications can be taken into account. Specifically, the networks have been populated with VoIP, video, and HTTP applications. Networks typically have to share multiple types of applications including voice, video, and data. We have seen in the literature review that most of the studies that exist in the literature have focused on studying capacity and performance of multimedia applications in isolation to other applications, that is, the networks are populated by the traffic that only belongs to one application type. These studies are therefore unable to capture the dynamics of realistic networking environments, because they do not take into account the mutual effects that various multimedia applications will have on each other, while sharing a single network resource. While there are some studies that have reported on multimedia performance while sharing the network resources with other applications, those studies are limited in their approach to setting up the applications and networking scenarios (e.g., the number and types of applications, and the analyses are limited).

15.4 Multimedia over Ad Hoc Networks: QoS Analysis

We now analyze video performance over wireless ad hoc networks using the four routing scheme as mentioned in Section 15.2; these are AODV, OLSR, GRP, and HCPR. We have simulated well over 500 different networking scenarios by varying the routing protocols, traffic intensities, mix of multimedia applications configurations, and numbers of nodes. Results show that different protocols act differently under different network conditions, and some protocols perform better at low traffic intensities while others give better performance for networks with higher numbers of nodes. In this section, we give a selection of scenarios and results from our simulation experiments, with one section devoted to video performance analysis. Our aim is to explore how the various protocols are affected by increasing the number of network nodes, traffic intensities, and the mix of multimedia traffic applications.

This section is organized as follows. Section 15.4.1 describes the performance analysis methodology including the simulation scenarios, traffic profiles and network topologies. Section 15.4.2 presents an analysis on the collective network behavior when the network is populated with multimedia applications. The results are presented separately for networks of different sizes, from a 5-node network up to a network containing 50 nodes. Section 15.4.3 gives a detailed depiction and discussion of video performance over ad hoc networks for different traffic profiles. In Section 15.4.4, we summarize and conclude with an overview of the QoS analysis for ad hoc networks presented in Sections 15.4.2 and 15.4.3. Subsequently, in Section 15.4.5, we analyze video performance over infrastructure networks. As expected, we will see that infrastructure networks are able to deliver much higher intensity of video.

15.4.1 Methodology for Analysis

We describe here our performance analysis methodology that includes: the four application profiles that we have created using a mix of multimedia (voice, video, and HTTP) applications (Section 15.4.1.1); the variation in the network size and geographical structure (Section 15.4.1.2); and the network topology of multimedia applications (Section 15.4.1.3).

15.4.1.1 Applications Traffic Profiles

In order to study the effect of how routing protocols are behaving with different traffic intensities, we have configured voice, video, and HTTP applications into four traffic intensities. These are listed in Table 15.1. Rows 1–4 give details for each traffic intensity beginning with Low (L) traffic at 100 kbps up to the High (H) intensity at 20 Mbps. Columns 1–4 list bit rate for the voice, HTTP, and video including the codec name for voice. Column 5 lists the total bit rate for each traffic intensity. The bit rates for each application also give the number of connections (this will be further explained in Section 15.4.1.3). For example, in the third row, which represents the details of Medium–High (MH) traffic, the first and second columns show the voice codec used and the total voice bit rate. GSM-FR codec is the codec used which has a bit rate of 12.3 kbps. The total voice bit rate is 52.3 which is a bit rate for two connections and each connection has two ways of voice.

Columns 3 and 4 show the total HTTP and video traffics which are 43.2 kbps and 9900 kbps = 9.9 Mbps, respectively. The last column is the total traffic populated in the network, including voice, HTTP, and video, which is 10,000 kbps = 10 Mbps. The last row shows the H intensity chosen at 20,000 kbps = 20 Mbps because this is usually the maximum achievable bandwidth for 802.11 g at 54 Mbps. As can be seen from the table, traffic populated were a mix of HTTP, voice, and video in all the four traffic intensities. Having a mix of data and multimedia applications in the same network was chosen to show how applications are performing with each other to simulate more realistic networking scenarios.

TABLE 15.1

Traffic Profiles and Their Respective Component Intensities and Attributes

15.4.1.2 Network Geography and Size

Varying the network size against fixed traffic intensities was intricate and required careful configuration, as it is difficult to compare the performance of two scenarios unless they are exactly the same, except one parameter, which is the one intended to study. In the case of this chapter, all parameters should be fixed and network sizes will be varied. The problem here is that increasing the number of nodes requires other parameters to change, which is the scenario of application sessions, that is which node is connected to which. For example, if a network has five nodes (A, B, C, D, and E) and there is a one video streaming from two nodes A and B. If number of nodes increases to 10, there will be 5 nodes added to the network. The challenge here is how to make use of the increase to 10 nodes, to assess that increase, with keeping the same application scenario. Since the goal of this study is to see the effect of number of hops increase when increasing number of nodes. In order to study the effects of increasing the number of hops on the network performance, nodes are fixed and distributed as follow. The study will start by having five nodes organized in one row and having distance of 200 m between each other. As number of nodes increases to 10, other row of 5 nodes will be added below and parallel to the first row. This is done consequently for 20, 30, 40, and 50 node scenarios. Figures 15.1 and 15.2 illustrate the geographical structure of a network with 5 and 20 nodes, respectively. The figures are explained further in the next section where we discuss applications topology.

FIGURE 15.1
The network geographical structure and applications topology (5 nodes).

15.4.1.3 Topologies

Having applications deployed similarly across network sizes is very important to study the effects of network size and hops increase. In all scenarios used in this study, there were two HTTP connections, two video streaming connections, and two voice connections. In case of five nodes, the pairs of connections of HTTP and video were initiated between nodes that are located in the edges of the network. The pair connections of Voice were between nodes that are located just to the next of the edge nodes. Figure 15.1 shows applications’ connections details for f5-node scenario. The red lines represent video, purples for HTTP, and green lines for voice. As the number of nodes increases to 10, the video connections will be between the two corners of the network. One video connection is between two facing corners which will be like a diagonal line. The Voice connection will be between nodes that are next to the two corners toward the middle of every row. The HTTP connections will be between the middle nodes in the first and second rows and between the two edges of the first row. The settings for 20-, 30-, 40-, and 50-node scenarios will follow the same patterns. Figure 15.2 depicts the applications topology for a network with 20 nodes.

15.4.2 Collective Multimedia Network Behavior against Application Profiles

Firstly, we study collective (overall) QoS behavior of ad hoc networks populated by multimedia applications (i.e., overall behavior of network hosting a range of video, voice, and HTTP applications). The performance is studied using four different routing protocols (HCPR, OLSR, AODV, and GRP), a set of four traffic intensities (100 kbps, 1 Mbps, 10 Mbps, and 20 Mbps) and six network sizes (network with 5, 10, 20, 30, 40, and 50 nodes). As mentioned earlier, in a later section, we will specifically look at the performance of vide traffic over ad hoc networks.

15.4.2.1 A 5-Node Network

Figures 15.3 and 15.4 show the overall DR and throughput for four routing protocols across four traffic intensities for a network containing five nodes. It can be observed that the two figures depict similar trends. For low traffic intensities (100 kbps and 1 Mbps), the network is able to sustain 100% throughput and DR. For higher traffic intensities, the network is unable to sustain the traffic load and starts dropping packets, resulting in lower DR and throughput, dropping to around 40% for 10 Mbps overall traffic and reaching to near 0 for 20 Mpbs. All four protocols show similar performance. AODV performs slightly better, we believe, due to its on-demand features: when other routing protocols consume part of their bandwidth by sending periodical routing messages, AODV uses the bandwidth to send the applications traffic. In the rest of this section, due to limited space, we will only present DR results as these show trends similar to throughput with some differences in the actual values.

FIGURE 15.2
The network geographical structure and applications topology (20 nodes).

FIGURE 15.3
Delivery ratio (DR) for 5-node network across four traffic intensities.

FIGURE 15.4
Throughput for 5-node network across the four traffic intensities.

15.4.2.2 A 10-Node Network

Figure 15.5 shows the total DR, as before, for four routing protocols across four traffic intensities, however, this time for a network containing 10 nodes. The figure depicts that DR for 100 kbps intensity varied from 81.8% for OLSR and GRP to 100% for HCPR protocol where AODV was in the middle of the two values with around 92.6% DR. As intensity was increased, HCPR was affected slightly with the increase by having 97.3% DR while GRP and OLSR affected badly to reach below 5.2% DR. AODV was not affected as OLSR and GRP but showed a big drop to reach around 54.4%. By increasing the traffic to 10 Mbps, all routing protocols have reached to below 3.2% DR with HCPR being the highest among the others.

FIGURE 15.5
Delivery ratio (DR) for 10-node network across the four traffic intensities.

FIGURE 15.6
Delivery ratio (DR) for 20-node network across the four traffic intensities.

15.4.2.3 A 20-Node Network

Figure 15.6 shows the total DR and throughput for a network with 20 nodes for 4 routing protocols across 4 traffic intensities. The figure depicts that different routing protocols have different DR when traffic was the lowest with values of 100%, 84.4%, 66.6%, and 53.8% for HCPR, AODV, GRP, and OLSR consequently. As traffic increased, all routing performances have fallen sharply except HCPR which has shown only 4% drop in performance. When traffic intensity reached 10 Mbps, all routing protocols’ performances went down to reach blow 1.2% with HCPR still being the highest among the other routing protocols. As traffic intensity reached the 20 Mbps, HCPR has gone to the lowest value of 0.3% while the other routing protocols has shown slight improvement of around 1% in comparison to the DR at 10 Mbps intensity, this is we believe due to the larger overhead that HCPR incurs due to clustering and reservations.

15.4.2.4 A 30-Node Network

Figure 15.7 shows the DR and throughput of 30 nodes scenarios for 4 routing protocols across 4 traffic intensities while network contains 30 nodes. The DR results show that all routing protocol performances were between 59.3% and 83.4% for 100 kbps traffic intensity, with HCPR in best performance position and OLSR and GRP at lowest position. As traffic increased to 1 Mbps, all routing protocols have shown a big drop in the performance to reach below 6.2%, and that trend continued to go down to reach below 1.4% when traffic was 10 Mbps. As traffic reached 20 Mbps, there has been slight improvement of around 1% for all routing protocols. The relatively poor DR performance for the 30-node network in comparison to 20-node network scenario is explained as follows. Firstly, this could be due to the MAC layer collisions caused by the increased number of transmissions attempts; this will in turn result in potential MAC buffer overflow caused by the packets arriving from higher layers. Secondly, it is because the increase in the number of network nodes has resulted into the increase in the geographical area of the network and hence a larger number of clusters. Additional nodes cause higher interference at the physical layer and add to the packet-dropping rate. Furthermore, a higher number of clusters can cause higher overhead and hence higher packet drops. These aspects of the HCPR scheme—that is, the relationship between the number of nodes, geographical area size, the traffic configuration, and the number/size of clusters—need further investigation. This will form our future work and further discussion on this topic is given in Section 15.5.

FIGURE 15.7
Delivery ratio (DR) for 30-node network across the four traffic intensities.

15.4.2.5 A 40-Node Network

Figure 15.8 shows the DR of a network with 40 nodes for 4 routing protocols across 4 traffic intensities. As can be observed from the figure that the results follow trends similar to the 30-node scenarios except the slight improvement of GRP over other routing protocols when traffic was 1 Mbps. The four routing protocols had DR between 63.2% and 83.4% when traffic was 100 kbps and between 1.2% and 1.5% when traffic was 20 Mbps.

FIGURE 15.8
Delivery ratio (DR) for 40-node network across the four traffic intensities.

FIGURE 15.9
Delivery ratio (DR) for 50-node network across the four traffic intensities.

15.4.2.6 A 50-Node Network

Figure 15.9 shows the DR of a network with 50 nodes for 4 routing protocols across 4 traffics intensities. As can be observed from the figure, the results follow almost the same trend as for 30-node and 40-node scenarios. The four routing protocols had DR between 52.8% and 82.4% when traffic was 100 kbps, which was slightly lower than scenarios of 40 nodes.

15.4.3 Video Performance

This section studies the performance of video applications over ad hoc networks. As before, the video application shared the network with voice and HTTP traffics. The study of video performance is for varying network sizes and a set of four traffic intensities; L, Medium–Low (ML), MH, and H. We plot results for DR, delay, and delay variation.

15.4.3.1 Traffic Profile: Low

Figure 15.10 depicts the performance of video DR when the network is populated with L overall traffic of 100 kbps and video traffic was around 68.5 kbps of the total traffic. The overall view of DR figure is a decrease with the increase of number of nodes except HCPR which showed excellent results at 10 and 20 nodes. That made HCPR to be the best achiever among routing protocols followed by AODV, which was in clear pattern and decreased gradually. OLSR dropped to around 34% when number of nodes was 20, rose at 30- and 40-node network sizes, and again dropped again at 50-node network size. GRP went down gradually until 30-node network size, and then started increasing slightly at 40- and 50-node scenarios.

Figure 15.11 shows the video packet end-to-end delay across network sizes when the networks are populated with L overall traffic of 100 kbps. The figure started by having L delay of around zero for all routing protocols and then started increasing at 30 nodes. AODV increased gradually from 30-node network size while HCPR increased sharply from 40-node network size. Although, HCPR was the worst at 50 nodes, it has the least lost ratio than others and it was the same for AODV as well. That means sometimes having a poor performance for one metric does not necessarily mean all metric performances are poor but it could indicate good performance in terms of other metrics, in this case DR was achieving good. Figure 15.12 shows the video delay variation across network sizes when the networks are populated with L overall traffic of 100 kbps. The variation was almost zero for all routing protocols across network sizes except for AODV and HCPR when number of nodes was 50 where the variation rose to just below 0.5 and 5 s consequently.

FIGURE 15.10
The video delivery ratio (DR) for L traffic across network sizes.

FIGURE 15.11
Video packet end-to-end delay for L traffic across network sizes.

FIGURE 15.12
Video packet delay variation for L traffic across network sizes.

15.4.3.2 Traffic Profile: Medium–Low

Figure 15.13 depicts the performance of video DR when the networks are populated with ML overall traffic of 1 Mbps, and video traffic was around 947 kbps of the total traffic. The overall view of DR figure is a decreasing trend with the increase of number of nodes. All routing protocols started with DR above 97.7% and then started decreasing with differences of the degree of the decrease. HCPR decrease was slightly and gradually to have 95.9% DR when number of nodes was 20 and then declined sharply to reach 4.5% at 30 nodes and kept under 5% for 50-node network size. AODV came after HCPR at 10 and 20 nodes. OLSR and GRP acted similarly by dropping to below 6% and remained like that except GRP where it has shown a slight improvement at 40 nodes. By comparing this figure with the overall traffic DR figure, we could conclude that video DR acted similar to the overall DR for L traffic profile.

FIGURE 15.13
Video delivery ratio (DR) for ML traffic across network sizes.

Figure 15.14 shows the video packet end-to-end delay across network sizes when the networks are populated with ML overall traffic of 1 Mbps. AODV was the only routing protocol for which packet end-to-end delay across all network sizes was recorded. It started like the other protocols by having low delay of around 0.03 s and kept the trend of low delay until it reached 30-node network size by having delay of around 0.09 s, and then rose sharply to reach 17 and 29 s at 40 and 50 nodes consequently. The delay recorded for HCPR was for 5-, 10-, and 20-node network sizes with delay between 0.02 and 0.07 s. GRP and OLSR recorded low delay at 5-node network size of around 0.02 s and at 40 nodes of around 60 and 76 s consequently. The reason for OPNET not to have recorded delays for some cases, we believe, is that the DR in those cases was zero (equivalently, it meant that the delay was infinite because the DR was zero).

Figure 15.15 shows the video delay variation across network sizes when the networks are populated with ML overall traffic of 1 Mbps. The delay variation results followed a trend similar to the end-to-end delay results; AODV was the only routing protocol that had values for this metric recorded across network sizes. It started by a delay value of near zero variation and kept it until 40-nodes network size where it increased to very high values and then went down at 50-node network size. As we said earlier, a low value of delay variation or delay is to be seen in the context of DR or throughput.

FIGURE 15.14
Video packet end-to-end delay for ML traffic across network sizes.

FIGURE 15.15
Video packet delay variation for ML traffic across network sizes.

15.4.3.3 Traffic Profile: Medium–High

Figure 15.16 depicts the performance of video DR when the networks are populated with MH overall traffic of 10 Mbps, and video intensity is 9.9 Mbps. As expected, the video DR decreases with increase in the number of nodes in the network. In general, the DR reaches to a near zero value for the network with 10 nodes. As the figure shows, AODV is slightly better than other routing protocols when the number of nodes in the network was 5: HCPR performed slightly better for 10- and 20-node networks; however, the difference in performance is negligible. Figure 15.17 shows video packet end-to-end delay results across the various network sizes when the networks are populated with MH overall traffic of 10 Mbps. HCPR had increasing trend moving from 5- to 20-node network size and did not record values for higher network sizes while others had discrete values. AODV had delays recorded between 5- and 10-node and between 40- and 50-node network sizes. GRP recorded one value at 5-node scenario while OLSR recorded values at 5- and 10-node network sizes. The delay of all routing protocols at 5-node network size was below 10 s which is the minimum values achieved at this intensity rate. Not having some values recorded at some points means that there is no video traffic being delivered to the destination.

FIGURE 15.16
Video delivery ratio (DR) for MH traffic across network sizes.

FIGURE 15.17
Video packet end-to-end delay for MH traffic across network sizes.

Figure 15.18 plots video delay variation results across network sizes when the networks are populated with MH overall traffic of 10 Mbps. The figure followed the same direction as the packet end-to-end delay for MH traffic. AODV showed a high variation at 5 nodes; a very high variation at 10- and 30-node network sizes; 0 variation at 40-node network size. HCPR variations started by having around 5 s at 5-node network size, above 20 s for 20- and 30-node network sizes, and then stopped recording for higher network sizes. OLSR and GRP had around 5 s variations at 5 nodes and then only OLSR had a variation of 18 s at 10 nodes.

FIGURE 15.18
Video packet delay variation for MH traffic across network sizes.

15.4.3.4 Traffic Profile: High

Figure 15.19 plots video DR results when the networks are populated with H overall traffic of 20 Mbps and video was 19.6 Mbps. The figure shows trends similar to the previous section, though the DR value is lower, obviously due to the higher traffic intensity. Figure 15.20 shows the video packet end-to-end delay. Some values were recorded for the protocols at 5- and 10-node network sizes, with very high delay between 7 and 11 s for 5-node network size. This high delay will not give a good interactive video streaming service but may be used for downloading noninteractive video streams. Figure 15.21 shows the video delay variation. The figure followed the same trend as the packet end-end-delay for H traffic, which was normal and expected. The values for recorded variations were from 5 to 120 s for 5- and 10-node network sizes, which are not acceptable for real-time video streaming. No values were reported by OPNET for network sizes greater than 10 nodes.

FIGURE 15.19
Video delivery ratio (DR) for High (H) traffic across network sizes.

FIGURE 15.20
Video packet end-to-end delay for H traffic across network sizes.

FIGURE 15.21
Video packet delay variation for H traffic across network sizes.

15.4.3.5 Video Performance for All Traffic Profiles: An Overview

We summarize here the results of video applications. A conclusion of each of the four traffic profiles is drawn using four metrics: DR, throughput, delay, and delay variation as follows. Generally, the performance of video DR was similar to overall DR performance because video traffic intensity was over 60% of the total traffic in case of L traffic and over 94% of ML, MH, and H intensities. As mentioned earlier, the video application shared the network with voice and HTTP traffic; therefore, there is more to the video behavior than the traffic intensity alone (e.g., the effects of VoIP applications on the network behavior). At L traffic intensity profile, all routing protocols achieved around 100% DR at 5-node network. HCPR was the only routing protocol that kept its high performance of 100% DR at 10- and 20-node networks. Packet end-to-end delay was below 0.25 s for all network sizes and below 0.02 when network sizes were between 5 and 20 nodes. At ML traffic intensity profile, all nodes had over 98% DR at 5-node network and then dropped to below 55% for larger networks except HCPR, which kept its high performance at 10- and 20-node networks to reach around 97.3% and 95.8%, respectively. In all cases where DR was above 95%, we had a maximum of 0.07 s delay and 0 delay variation. At MH and H traffic intensities, none of the routing protocols achieved acceptable DR at any network sizes. In conclusion, HCPR had the best performance in L and ML traffic intensities.

15.4.4 Ad Hoc Networks: Performance Overview

In this section (Section 15.4), by now, we have presented an extensive analysis of the HCPR performance for multimedia delivery over ad hoc networks. A number of scenarios were configured by varying different routing protocols, different intensities, and different numbers of nodes. It showed how protocols are affected by increasing the number of nodes and traffic intensities and how HCPR was performing in comparison to other routing protocols (AODV, GRP, and OLSR) under different scenario settings. Network sizes were varied by the number of nodes from 5-node to 50-node networks. In addition, traffic intensities were set in four main intensity profiles, L, ML, MH, and H. Each intensity profile was a mix of three multimedia applications video, voice, and HTTP. We have studied the overall performance by fixing the network size and looked into as to how networks were behaving under different intensity loads and mix of applications. That was followed by a performance analysis of multimedia applications by fixing traffic intensity and varying network sizes. While studying the effects of varying traffic intensities across network sizes, HCPR was the only routing protocol that showed a good overall performance at L and ML traffic intensities for 10- and 20-node networks. When studying the effect of varying network sizes for each traffic intensity profile, the collective network performance and video performance for HCPR outperformed other routing protocols at 10- and 20-node networks for L and ML traffic profiles. The high bitrate of video and HCPR performer have resulted in having lower performance of voice especially at MH and H traffic intensities. The HTTP results also demonstrated that HCPR was the best performer at L and ML intensities, for higher traffic intensities, HCPR also outperformed other routing protocols at 30- and 40-network sizes.

15.4.5 Infrastructure Wireless Networks: QoS Analysis

We now present a QoS analysis of the results obtained through simulating video applications over an infrastructure wireless network. We will see that, in contrast to the results for ad hoc networks presented in the earlier part of this section (Section 15.4), the networks will be able to support much larger video intensity. The input parameters used in these simulations include three different voice codecs and three different traffic intensity levels for each of video and HTTP application. In addition, there are 12 different variation levels in (latency, packet discard ratio) pair for metropolitan area network (MAN). We have performed well over 100 network simulations by using different input configurations obtained through variations in these input parameters’ values.

The network topology and architecture is described in Section 15.4.5.1, followed by a description of multimedia traffic, including voice, video, and data used in the QoS analysis in Section 15.4.5.2. The QoS results are discussed in Section 15.4.5.3. As before, the network and applications have been simulated using the OPNET simulator. We will follow the same performance analysis methodology as described in Section 15.4.1, that is, the network is populated with a mix of video, voice, and data applications.

15.4.5.1 The Network Architecture

We are interested in exploring video application performance in MAN environments where the two end hosts access the network through 802.11 Wireless LANs. Specifically, we are interested in investigating video performance over 802.11 networks under realistic scenarios where the network bandwidth is being shared by other applications including voice and HTTP traffic. Networks are usually heavily loaded with video, data, and other traffic. Hence, in particular, we are interested in investigating multimedia performance over a range of network loads. The structure of the network which we have considered here is described as follows. The network consists of two WLANs which are connected to a common MAN via 1 Gbps Ethernet. We have used 802.11 g (54 Mbps) for each of the two WLANs. For the MAN, we have used an abstract simulation model (a cloud) which allows us to set different parameters for the network including packet discard ratio and latency. A total of 10 nodes are associated to each of the WLANs. These 10 nodes are randomly located within an area of 350×350 m² around each of the two WLANs. The 20 nodes interact with each other across the network using different applications (see the next section).

TABLE 15.2

Traffic Profiles for Wireless Infrastructure Networks (kbps)

	Low (L)	Medium (M)	High (H)
Video	21,600	2800	33,600
Voice	G723.1 (5.3)	GSM-FR (13)	G.711 (64)
HTTP	167	168	169
Total	21,800	28,300	34,400

15.4.5.2 Applications

We populate the network with varying intensity levels of video, HTTP, and voice applications. We use three different traffic intensity levels for each of the video, HTTP, and voice applications; these are listed in Table 15.2. In a similar approach to the one taken in Section 15.4.1.1, the traffic intensity levels are divided into “Low,” “High,” and “Medium.” Columns 2–4 outline the details for each application specific to the three traffic intensity levels. For example, Column 3 lists the voice codec used for the “Medium” intensity traffic case is GSM-FR with 13 kbps bitrate; the video and HTTP traffic in the “Medium” intensity case are 28 Mbps and 168 kbps, with a net traffic of 28.3 Mbps. Similarly, Column 2 and 4 give the details relevant to the “Low” and “High” traffic intensity levels. In each of these cases, 10 nodes from 1 WLAN are connected to the 10 nodes in the other WLAN. Each node in a WLAN is connected to a node in the other WLAN, and hence there are a total of 30 connections, 10 each for HTTP, video, and voice.

15.4.5.3 QoS Analysis

Having described the network architecture and applications set up, we are now ready to analyze QoS for wireless infrastructure networks connected through a MAN. First, we discuss the network results while keeping constant the settings for the MAN connecting the WLANs. Secondly, we analyze network QoS against variations in the MAN.

FIGURE 15.22
(See color insert.) Video QoS—traffic delivered and throughput.

Figure 15.22 plots data about video traffic in the network. The x-axis represents intensity of the total traffic that populates the network. There are two y-axes: the one on the right side represents the video traffic sent and delivered (received) through bar graphs and the other on the left side represents the video throughput in percentage through the line plot. Note that the figure plots the video statistics where the network is also populated with voice and data applications as detailed earlier. We note in the figure that the network is able to deliver 18.6 MB of video per second out of the 21 MB of the transmitted video content, resulting in 88.3% throughput. This value of throughput is maintained for up to 21.6 MB of transmitted video, and then the throughput starts dropping until it reaches to the lowest value of 6.5% for 33.6 MB of video transmitted on the network. The total theoretical capacity of the WLAN is 54 Mbps and therefore it is expected that the network will get saturated at some threshold point and will start dropping the packets in great numbers.

Figure 15.23 plots the network traffic intensity in Mbps for all the network applications as well as the total throughput for the entire network. Note that Figure 15.22 illustrates the video performance alone. There are a total of five plots in the figure; three of these plots represent the traffic produced by the three applications (voice, video, and HTTP), the fourth gives the total traffic intensity, and the fifth gives the throughput obtained against varying network traffic intensity. The total network traffic intensity and total network throughput are also visible in the figure. We note in the graph that initially (i.e., from the left, for lower values of network intensity), the traffic intensity and the throughput bar graphs are of equal heights showing that all the traffic sent to the network is delivered with 100% throughput. However, as the traffic reaches approximately 21 Mbps, the difference in the heights of the two bar graphs become visible with the throughput reaching its lowest value of 2.4 Mbps against the total input traffic of 33.6 Mbps. As for Figure 15.22, this behavior is expected due to the capacity limits of the wireless network.

FIGURE 15.23
(See color insert.) Network traffic components—intensities and throughput.

Figures 15.24 and 15.25 depict end-to-end delay characteristics for (infrastructure) wireless networks connected through a MAN. We now make some variations in the metropolitan network environment which connects the two WLANs. Specifically, we vary the input parameter values for the metro network latency and packet discard ratio. The OPNET software allows us to configure various parameters of the MAN environment through its “Cloud” model. Figure 15.24 plots the average end-to-end delay of the network against variations in traffic intensity and packet discard ratio. We note that the delay increases significantly with increase in the traffic intensity; however, it remains relatively almost constant against changes in the packet discard ratio. Figure 15.25 plots the end-to-end delay against traffic intensity and network latency. As is the case for Figure 15.24, the delay increases with the increase in traffic intensity but remains relatively almost constant against metro network latency. This is due to the fact that the change in the network latency (in MAN environments) is quite low compared to the voice delay caused by the limitations of the WLAN throughput capacity. A detailed analysis of multimedia applications over wireless networks connected within a MAN is presented in Mehmood et al. (2011) with a focus on VoIP applications.

FIGURE 15.24
(See color insert.) Average network delay against MAN packet discard ratio and network traffic.

FIGURE 15.25
(See color insert.) The average network delay against MAN latency and network traffic intensity.

15.5 Conclusions and Future Work

Multimedia applications continue to drive convergence in the telecommunications industry. Video traffic is to become 90% of the overall global network traffic by 2015. People demand anytime, anywhere mobility giving rise to heterogamous networks comprising a range of fixed, wireless, mobile, and ad hoc networks. Multimedia applications pose stringent requirements on networks while ad hoc networks are inherently limited due to their autonomic and multi-hop nature and the lack or scarcity of infrastructure. Supporting multimedia applications over ad hoc networks is a challenge for the industry. The broad aim of our research is to address the challenges associated with the design and analysis of ad hoc networks supporting multimedia applications with scalable QoS.

The aim of this chapter was to present an overview of our work on multimedia wireless networks. We made three contributions in this chapter. Firstly, in Section 15.3, an extensive review of the literature on multimedia ad hoc network design and QoS analysis was presented. Secondly, in Section 15.4, a detailed analysis of multimedia applications over infrastructure and ad hoc networks was presented. Four routing protocols were used in this analysis and over 500 networking and applications scenarios were simulated and it was demonstrated that the HCPR-enabled ad hoc network outperforms the well-known routing schemes, in particular for relatively large networks and high QoS network loads. These results show promising performance because many QoS schemes do work for small networks and low network loads but are unable to sustain performance for large networks and high QoS loads. Thirdly, a detailed end-to-end QoS analysis for multimedia applications was presented over wireless infrastructure networks connected within a MAN environment. All the network analyses presented in this chapter are based on simulations using the OPNET simulator. The HCPR scheme is implemented as an extension (module) to the OPNET simulation software.

We had also contended in this chapter that the vast majority of networks are populated with multimedia traffic, as opposed to video alone, and therefore our analysis in this chapter have intentionally included voice and data. Nevertheless, we have also presented network analysis with focus on video application alone. In this context, a contribution of our work lies in the analysis methodology that we have developed for configuring multimedia applications and networking scenarios. The approach was to simulate and evaluate realistic multimedia applications scenarios for networks with a good number and type of applications—VoIP, video, and HTTP, both elastic and nonelastic—so that the mutual effects of network applications can be taken into account. The work on analysis methodology can be extended and enhanced in the future by using additional types of multimedia traffic models (such as video codecs and models, additional VoIP models, etc.) and network configurations. We would like to formalize our analysis methodology by finding good and bad practices for networks and applications setup found in realistic environments.

Acknowledgment

We are thankful to all the reviewers for their time, very useful recommendations, and their comments for helping us to improve this chapter.

References

Alazawi, Z., Altowaijri, S., Mehmood, R., and Abdljabar, M. B. 2011. Intelligent disaster management system based on cloud-enabled vehicular networks. In: 11th International Conference on Proceedings of the ITS Telecommunications (ITST), August 23–25, 2011, St. Petersburg, Russia, pp.361–368.

Alturki, R. 2011. Multimedia ad hoc networks: Design, QoS, routing and analysis. PhD thesis, Swansea University, Swansea, U.K.

Alturki, R. and Mehmood, R. 2008. Multimedia ad hoc networks: Performance analysis. In: EMS ‘08. Second UKSIM European Symposium on Computer Modeling and Simulation, IEEE Computer Society 2008, September 8–10, 2008, Liverpool, England, UK, pp. 561–566.

Alturki, R. and Mehmood, R. 2012. Cross-layer multimedia QoS and provisioning over ad hoc networks. In: Rashvand, H. F. and Kavian, Y. S. (eds.) Using Cross-Layer Techniques for Communication Systems: Techniques and Applications. IGI Global, Hershey, PA, pp. 460–499.

Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., and Weiss, W. 1998. An Architecture for Differentiated Services. RFC:2475. IETF Network Working Group.

Braden, R., Clark, D., and Shenker, S. 1994. Integrated Services in the Internet Architecture: an Overview. RFC 1633. IETF Network Working Group.

Broch, J., Maltz, D. A., Johnson, D. B., Hu, Y.-C., and Jetcheva, J. 1998. A performance comparison of multi-hop wireless ad hoc network routing protocols. In: Osborne, W. P. and Moghe, D. (eds.), Proceedings of the 4th Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom ‘98), ACM, New York, NY, 1998, pp.85–97.

Camp, T., Boleng, J. and Davies, V. 2002. A survey of mobility models for ad hoc network research. Wireless Communications & Mobile Computing (WCMC): Special issue on Mobile Ad Hoc Networking: Research, Trends and Applications, 2, 483–502.

Chatzistavros, E. and Stamatelos, G. 2009. Comparative performance evaluation of routing algorithms in IEEE 802.11 ad hoc networks. In: Proceedings of the 16th International Conference on Telecommunications, May 25–27, 2009, Marrakech, Morocco: IEEE Press, pp.19–24.

Cisco Systems, I. 2006. Understanding Jitter in Packet Voice Networks (Cisco IOS Platforms). Available: http:/www.cisco.com/en/US/tech/tk652/tk698/technologies_tech_note09186a00800945df.shtml (Accessed October 5, 2010).

Clausen, T. and Jacquet, P. 2003. Optimized link state routing protocol (OLSR). RFC: 3626. IETF Network Working Group.

Gharavi, H. and Ban, K. 2003. Cross-layer feedback control for video communications via mobile ad-hoc networks. In: IEEE 58th Vehicular Technology Conference, VTC 2003-Fall, Orlando, FL, Vol. 5, October 6–9, 2003, pp. 2941–2945.

Gottron, C., Konig, A., Hollick, M., Bergstrasser, S., Hildebrandt, T., and Steinmetz, R. 2009. Quality of experience of voice communication in large-scale mobile ad hoc networks. In: Wireless Days (WD), 2nd IFIP, December 15–17, 2009, Paris, France, pp. 248–253.

Hofmann, P., An, C., Loyola, L., and Aad, I. 2007. Analysis of UDP, TCP and voice performance in IEEE 802.11b multihop networks. In: 13th European Wireless Conference, April 1–4, 2007, Paris, France.

Hong, X., Gerla, M., Guangyu, P., and Ching-Chuan, C. 1999. A group mobility model for ad hoc wireless networks. In: Proceedings of the 2nd ACM International Workshop on Modeling, Analysis and Simulation of Wireless and Mobile Systems. Seattle, WA. New York: ACM Press, pp.53–60.

Hongqi, Z., Jiying, Z., and Yang, O. 2008. Adaptive rate control for VoIP in wireless ad hoc networks. In: IEEE International Conference on Communications, ICC’08, May 19–23, 2008, Beijing, China, pp.3166–3170.

IEEE 802.11 Working Group (2007). IEEE 802.11-2007—Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications.

Jeong, Y., Kakumanu, S., Tsao, C.-L., and Sivakumar, R. 2009. VoIP over Wi-Fi networks: performance analysis and acceleration algorithms. Mobile Networks and Applications, 14, 523–538.

Kumar, S., Rathy, R. K., and Pandey, D. 2009. Traffic pattern based performance comparison of two reactive routing protocols for Ad hoc networks using NS2. In: 2nd IEEE International Conference on Computer Science and Information Technology, ICCSIT 2009, August 8–11, 2009, Beijing, China, pp.369–373.

Layuan, L., Chunlin, L., and Peiyan, Y. 2007. Performance evaluation and simulations of routing protocols in ad hoc networks. Computer Communications, 30, 1890–1898.

Lee, G. and Song, H. 2010. Cross layer optimized video streaming based on IEEE 802.11 multi-rate over multi-hop mobile ad hoc networks. Mobile Networks and Applications, 15, 652–663.

Li, X. and Cuthbert, L. 2005. Optimal QoS mechanism: integrating multipath routing, DiffServ and distributed traffic control in mobile ad hoc networks. In: Jia, X., Wu, J. and He, Y. (eds.), Proceedings of the First International Conference on Mobile Ad-hoc and Sensor Networks (MSN’05). Berlin, Germany: Springer, pp.560–569.

Li, J., Jannotti, J., Couto, D. S. J. D., Karger, D. R. and Morris, R. 2000. A scalable location service for geographic ad hoc routing. In: Proceedings of the 6th Annual International Conference on Mobile Computing and Networking. Boston, MA. New York: ACM.

Lidong, Z. and Haas, Z. J. 1999. Securing ad hoc networks. IEEE Network, 13, 24–30.

Liu, C. and Kaiser, J. 2003. A survey of mobile ad hoc network routing protocols. Department of Computer Structures, University of Ulm, Ulm, Germany.

Mehmood, R. and Alturki, R. 2011. A scalable multimedia QoS architecture for ad hoc networks. Multimedia Tools and Applications, 54(3), 551–568.

Mehmood, R., Alturki, R., and Zeadally, S. 2011. Multimedia applications over metropolitan area networks (MANs). Journal of Network and Computer Applications, 34, 1518–1529.

Melodia, T. and Akyildiz, I. F. 2010. Cross-layer QoS-aware communication for ultra wide band wireless multimedia sensor networks. IEEE Journal on Selected Areas in Communications, 28, 653–663.

Mohapatra, P., Chao, G., and Jian, L. 2004. Group communications in mobile ad hoc networks. Computer, 37, 52–59.

de Morais Cordeiro, C., Gossain, H., and Agrawal, D.P. 2003. Multicast over wireless mobile ad hoc networks: Present and future directions. IEEE Network, 17, 52–59.

Morgan, Y. L. and Kunz, T. 2005. A proposal for an ad-hoc network QoS gateway. In: WiMob’2005, August 2005, Montreal, Canada, Vol.3, pp. 221–228.

Mundarath, J. C., Ramanathan, P., and Veen, B. D. V. 2009. A quality of service aware cross-layer approach for wireless ad hoc networks with smart antennas. Ad Hoc Networking, 7, 891–903.

Murthy, C. S. R. and Manoj, B. S. 2004. Ad Hoc Wireless Networks: Architectures and Protocols. Upper Saddle River, NJ: Prentice Hall, PTR.

Ng, P. C. and Liew, S. C. 2007. Throughput analysis of IEEE802.11 multi-hop ad hoc networks. IEEE/ACM Transactions on Networking, 15, 309–322.

O’Hara, B. and Petrick, A. 2005. The IEEE 802.11 Handbook: A Designer’s Companion, 2nd edn. New York: Standards Information Network, IEEE Press.

OPNET Technologies Inc. 2008. OPNET Modeler-Educational Version, 14.5 ed.

Perkins, C., Belding-Royer, E., and Das, S. 2003. Ad hoc on-demand distance vector (AODV) routing. RFC: 3561. IETF Network Working Group.

Pompili, D. and Akyildiz, I. F. 2010. A multimedia cross-layer protocol for underwater acoustic sensor networks. IEEE Transactions on Wireless Communications, 9, 2924–2933.

Pucha, H., Das, S. M., and Hu, Y. C. 2007. The performance impact of traffic patterns on routing protocols in mobile ad hoc networks. Computer Networks, 51(12), 3595–3616.

Qadri, N. N. and Liotta, A. 2010. Analysis of pervasive mobile ad hoc routing protocols. In: Aboul-Ella, H., Jemal, A., Ajith, A., Hani, H. (eds.), Pervasive Computing, Computer Communications and Networks, Part 4, pp. 433–453.

Ramanathan, R. and Redi, J. 2002. A brief overview of ad hoc networks: challenges and directions. IEEE Communications Magazine, 40, 20–22.

Rosen, E., Viswanathan, A. and Callon, R. 2001. Multiprotocol label switching architecture. RFC 3031. IETF Network Working Group.

Santos, N. P. 2009. Voice traffic over mobile ad hoc networks: A performance analysis of the optimized link state routing protocol. Master’s thesis, Air Force Institute of Technology, Wright-Patterson AFB, OH.

Satyanarayanan, M. 2010. Mobile computing: The next decade. In: Proceedings of the 1st ACM Workshop on Mobile Cloud Computing and Services: Social Networks and Beyond, (MCS ‘10), ACM, New York, NY, Article 5, p.6.

Setton, E., Taesang, Y., Xiaoqing, Z., Goldsmith, A., and Girod, B. 2005. Cross-layer design of ad hoc networks for real-time video streaming. IEEE Wireless Communications Magazine, 12, 59–65.

Sondi, P., Gantsou, D. and Lecomte, S. 2010. Performance evaluation of multimedia applications over an OLSR-based mobile ad hoc network using OPNET. In: 12th International Conference on Computer Modelling and Simulation (UKSim), March 24–26, 2010, Cambridge, U. K., pp.567–572.

Takagi, H. and Kleinrock, L. 1984. Optimal transmission ranges for randomly distributed packet radio terminals. IEEE Transactions on Communications, 32, 246–257.

Utsu, K., Chow, C. and Ishii, H. 2010. A study on video performance of multipoint-to-point video streaming with multiple description coding over ad hoc networks. Electrical Engineering in Japan, 170, 43–50.

Xue, Q. and Ganz, A. 2003. Ad hoc QoS on-demand routing (AQOR) in mobile ad hoc networks. Journal of Parallel and Distributed Computing, 63, 154–165.

* See, for example the various multimedia services offered by TrafficLand, Inc (http://www.trafficland.com/). They provide services based on aggregation and delivery of live traffic video over the Internet and on TV.

† An ad hoc network is usually defined as a collection of nodes dynamically forming a network without the use of any existing network infrastructure or centralized administration.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for 15. Video QoS Analysis over Wi-Fi Networks

Create new playlist

Sign In

Sign Up

Table of Contents for
15. Video QoS Analysis over Wi-Fi Networks