6
MOBILE TV USING 3G TECHNOLOGIES
The significant problems we face can not be solved at the same level of thinking we were at, when we created them.
—Albert Einstein
It began with the 2.5G networks such as GPRS, EDGE, and cdmaOne in the late 1990s as a service for streaming of short clips. The operators had upgraded the networks from pure voice to being data capable. Users could set up data calls using circuit-switched connections or always-on packet-switched GPRS connections. cdmaOne and GPRS users had always-on connectivity using packet-switched connections. Wireless Application Protocols (WAP) was formalized and was intended to be the protocol of choice for accessing wireless applications over the air. However, in the initial period at least, the data usage of the networks was limited. Internet access, though possible, had limited attraction owing to the tiny screens and limitations of keypads and indeed of the cell phones themselves. Operators keen to derive maximum benefit from the networks paralleled the Internet, where the video streaming services had already become widespread as had video calling, using protocols such as H.261 on the fixed-line telecommunications networks.
The availability of highly compressed video clips under the new compression algorithms such as Windows Media or MPEG-4 and their progressive standardization under the Third Generation Partnership Project (3GPP) forum as well as the ITU and other agencies made it advantageous for the mobile operators having 2G and 2.5G networks to leverage on the capacities for data in their networks and provide video clips. Many GSM-, GPRS-, and CDMA-based networks started offering the clip download services as well as limited video streaming. This was in no small measure facilitated by the increasing power of mobile phones for handling multimedia applications such as audiovisual content. The initial video streaming services were limited to small clips of, say, 30 sec and low frame rates of 15 fps. As the availability of handsets and the usage grew the 2.5G networks were already straining the limits of their capacity in terms of streaming or downloading video to a large number of users, and the limitations were quite obvious in the form of frozen frames and interrupted video viewing, as average bit rates on 2/2.5G connections averaged 40–50 kbps. This brought the focus back on the 3G networks, which were designed to have a greater capability for data. When we speak about mobile TV services being provided over 3G networks, we in effect mean the 3G networks as well as the enhancements developed, such as Multimedia Broadcast and Multicast Services (MBMS), 1× Evolution Data Optimized (1×EV-DO), and High-Speed Downlink Packet Data Access (HSDPA). These enhancements were driven by the need to make available mobile TV services to ever-increasing numbers of users within the same cellular networks.
Today video clip streaming services as well as live video are widely available for a broad range of content, such as news headlines, weather, favorite shots in sports, and cartoons. Mobile versions of popular programs were the first choice for such implementations, together with content and programs designed specifically for mobile TV. For example:
• Mobile ESPN, a wireless service, was launched for sports fans.
• The GoTV network offers content from ABC and Fox Sports, as well as original programming.
• Verizon V CAST is a video clip download service offered over Verizon’s EV-DO and CDMA2000 networks.
• Sprint TV Live! provided by Sprint over its PCS Vision network offers a number of channels of continuously streamed content, most of them live.
Operators in Europe, Asia, and Latin America all offer such services. Reliance Infocom in India offers a near-live video streaming service comprising news channels over its CDMA2000 network. HBO content is available over Cingular and other networks in the United States.
The Universal Mobile Telecommunication System (UMTS) networks under the IMT2000 framework were primarily designed to provide high user bit rates for mobile customers. 3G-UMTS provides circuit-switched connections up to 384 kbps and packet-switched connections up to 2 Mbps. This is achieved by using 5 MHz carriers, improved radio interfaces, and core architectures.
We need also to recognize that the IMT2000 framework, which was formulated in the 1990s, is not the ultimate solution for providing live-TV-type applications to an unlimited user base owing to the resource constraints within the network.
6.2 WHAT ARE TV SERVICES OVER MOBILE NETWORKS?
Video services over mobile networks (including live TV) are provided by streaming the video and audio over the networks in a manner very similar to streaming over the Internet. There are certain differences, however, in the way in which video is streamed over the mobile networks (as opposed to the Internet), which essentially relate to the characteristics of the mobile networks.
Streaming as a method of transferring video, audio files, or live data has the advantage that the user need not await the full file download and can commence viewing the content while receiving the data. However, as experience with video streaming over Internet has shown, the streaming service quality is subject to sustained rates of data transfer over the network. Hence the quality of streamed video and the number of users that can use such services are dependent on the underlying mobile network, which is important to understand. The limitations of unicasting live TV content has led to broadcast and multicast technologies such as MBMS.
Mobile TV services are facilitated by the use of common standards for the file formats and coding of audio and video as formalized under the 3G Partnership Project and the 3GPP Packet Streaming Standards (3GPP-PSS).
In order to carry video and audio representing multimedia files or live streaming TV the following are the requirements that need to be met by cellular mobile networks:
Mobile networks should be able to establish calls, identified as video calls, in which the live video of the caller is associated with the call and vice versa. For live streaming video delivered by a packet-switching service, there needs to be a well-defined protocol as well, which identifies the nature of call. Hence, the first requirement is to have protocols standardized and agreed on for calling, answering, and establishing a video call or video streaming. These protocols need to be followed identically across networks so that the calls can be established between users on different networks. Having well-defined protocols also helps the handset manufacturers to deliver phones that can work identically on various networks. The procedures for setting up calls as well as the packet-switched streaming have been formalized under the 3G-324M recommendations for video calls and the 3GPP-PSS for video streaming, respectively.
The networks must have standards for encoding of video and audio defined for different applications, such as video calling or video streaming. Ideally the protocols would use high-efficiency compression algorithms such as MPEG-4 or H.264 in order to reduce bandwidth requirements for encoded video and audio. With small screen sizes it is also possible to use simple profiles for video, which do not require coding of a large number of objects. It is common to use visual simple profile for video, which has been formalized under the 3GPP.
The networks must have an adequate data rate available for uninterrupted transfer of video frames. In practice the video throughput data rates provided by the connection may vary, but on average should be maintained above certain minimum rates based on video parameters (for example, 64 kbps or above for video calls using QCIF encoding). It is possible to carry video on 2G, 2.5G, or 3G networks provided the minimum data rates can be maintained. In practice only 2.5G networks can provide any significant transmissions of streaming video and 3G networks are needed to provide satisfactory services.
In this chapter we look at the protocols to set up and initiate video data transfer and release such calls. We also look at the 3GPP packet-switching protocols, which make multimedia data transfer possible, and the 3GPP standards for handling video and audio. Mobile networks are characterized by the unicast or multicast capabilities for carrying video and we look at these modes of video delivery as well.
6.3 OVERVIEW OF CELLULAR NETWORK CAPABILITIES FOR CARRYING MOBILE TV
We begin by taking a look at the data capabilities of 2G, 2.5G, and 3G networks and follow this by the handling of various mobile TV and multimedia applications (such as video telephony) using mobile networks.
6.3.1 Data Services in 2G and 2.5G Networks
Data services in the GSM networks are provided by a circuit-switched data service. This requires the mobile to establish the data connection and after the data call has been established, the user has exclusive use of the data slot at 9.6 kbps. Higher data rates can be achieved by using enhanced coding techniques, which can raise the data rates to 14.4 kbps per slot. Multiple slot utilization is possible, which can permit the rates to go up to 38.4 kbps with the use of four slots (or 57.6 kbps using enhanced coding). The use of circuit-switched data for Internet-type services such as browsing is not an ideal mode of usage as the bandwidth of the synchronous connection is dedicated to the user who has set up the data call and there is no mechanism to share it with other users.
The General Packet Radio Service (GPRS) is an add-on feature to the existing GSM networks that provides the facility to set up and transfer packet-switched data in a shared mode among all users through a high-speed packet-switched IP network. The GPRS add-on functions are achieved by using the Gateway GPRS support node (GGSN) and Serving GPRS support nodes (SGSN).
In GPRS there are four different coding schemes, termed CS-1 to CS-4, which are used depending on the radio conditions. Coding scheme CS-1 provides a bit rate of 9.05 kbps and is used under the worst radio conditions. The highest coding scheme, CS-4, can provide a data rate of 21.4 kbps per slot, but under the best radio conditions, as the scheme is without any error correction facility.
Circuit-Switched Data Rates in GSM Networks
The most commonly used scheme is CS-2, with which a data rate of 53.6 kbps can be achieved using four time slots (Table 6-1).
The Enhanced Data Rates for Global Evolution (EDGE) is an improvement over the GPRS networks that refined the air interface between the mobile station and the base station. EDGE uses GMSK or 8PSK modulation depending on radio conditions. Through EDGE a theoretical total maximum rate of 473 kbps can be achieved using all eight time slots.
6.3.2 Data Capabilities of 3G Networks
The 2.5G and 2.75G networks such as GPRS, CDMA 1×, and EDGE were brought forth by the 2G operators in order to make applications such as mobile data and streaming video/audio feasible over their networks. However, because of the widespread use of data and other services the theoretical peak download rates could rarely be achieved. Instead the users experienced data throughputs of 20 kbps on average on the GPRS networks and 40–50 kbps on the EDGE networks. This performance can further degrade over the peak hours. This meant that any video longer than 30–60 sec is onerous to download under normal conditions.
Live video carriage requires at least 100–128 kbps with 15 fps and QCIF resolution with MPEG-4 coding. This is obviously not possible in 2.5G networks, and 3G networks became the medium for offering such services (Fig. 6-1).
FIGURE 6-1 Mobile Data Evolution toward Higher Data Rates (3G Systems)
6.3.3 Classification of 3G Networks
The network evolution from voice-oriented networks such as 2G (GSM or cdmaOne) to 3G has taken place in two branches, i.e., those involving the CDMA networks and those involving the GSM networks. The GSM networks evolved to GPRS/EDGE, which are 2G (some times called 2.5G) technologies, and are finally evolving to 3G as per the UMTS framework. The UMTS framework prescribes the air interface standard of WCDMA for the 3G services. The CDMA networks on the other hand evolved from IS95A (capable of 14.4 kbps data) to IS-95B (64 kbps) and then to CDMA2000, which is a 3G standard. Higher evolutions have followed a path of multiple 1.25 MHz carriers, i.e., CDMA2000 1×, CDMA2000 3×, and CDMA2000 6×, in order to meet the demands of real-time mobile TV as well as other applications. Both the technology lines are within the IMT2000 framework. However, while the WCDMA standard (UMTS) is a direct spread technology, the CDMA2000 standards have grown using the technology of multicarrier (CDMA2000) or time division duplex (TDD) (UTRA TDD and TD-SCDMA) (Fig. 6-2).
FIGURE 6-2 Wireless Technology and 3G Evolution Overview
CDMA (CDMA2000) technologies
• 2G: CDMA (IS-95A, IS-95B)
• 2.5G: 1×RTT
• 3G: 3×RTT
• 3G: EV-DO
• Enhanced 3G: EV-DO and EV-DO revisions A and B
GSM-based technologies
• 2G: GSM
• 2.5G/“2.5G +”: GSM/GPRS/EDGE
• 3G: WCDMA (UMTS)
• 3G MBMS
• Enhanced 3G (3.5G): HSDPA
FIGURE 6-3 3G Services in the United States—CDMA and UMTS
An example of how the two separate planes of GSM-evolved networks (UMTS or WCDMA) and CDMA-evolved networks exist in the United States is shown in Fig. 6-3.
3G networks and data services are a fact of life today, with over 115 million 3G customers using a range of data services by mid-2006. The mobile world, with 2 billion users in 2006, is on a growth path with over 30% growth on average. The 2G growth is being driven by markets in India, China, Russia, and other countries. If the trend observed in evolved networks is any indication the move toward 3G networks as well as 3G+ -evolved capabilities is only to be expected in the near future.
6.3.4 FOMA—First 3G Service from Japan
Japan was the first country to introduce the 3G services, with the launch of FOMA in 2002 (Fig. 6-4). The service, which had rich applications driving the underlying technology of 3G, had over 30 million subscribers in 2006. Japan has had a history of successful launches of interactive mobile services since NTT DoCoMo launched its i-mode service in 1999, which essentially brought the Internet to phones in Japan. The service proved very popular with its Yellow “I” button, which gave mobile users the Internet access option with a number of predefined application menus. The services proved so popular that by 2004 one-third of the Japanese, over 44 million, were using i-mode services. The services later gave way to FOMA, a new 3G service of NTT DoCoMo (Fig. 6-5).
FIGURE 6-4 FOMA Services—Japan (Picture Courtesy NTT DoCoMo)
The reason for the success of the i-mode services was believed to be ready applications from participating companies such as rail and air ticket booking, e-mail, music download, and shopping.
6.3.5 MobiTV
MobiTV, which is available in the United States (Cingular, Alltel, and regional carriers), Canada (Bell Canada, Rogers, and TELUS Mobility), the United Kingdom (Orange UK and 3), and other countries, provides over 40 channels as live TV content and also video on demand. In early 2006 the service was available at a flat rate of $9.99 U.S. per month, which was charged over and above the data rate plan. The channels available include MSNBC, ABC News Now, CNN, Fox News, Fox Sports, ESPN 3GTV, CNBC, CSPAN, the Discovery Channel, TLC, and others. Over 50 handsets can receive MobiTV telecasts. The technology of MobiTV was developed by Idetic.
FIGURE 6-5 Evolution of 3G Mobile Phone Services in Japan (Courtesy NTT DoCoMo)
The MobiTV service streams video on the data channel and the quality depends on a number of factors, including the network conditions and the phone selected.
In the United States the Cingular Wireless network is based on the WCDMA (UMTS) technology, providing average speeds of 200– 320 kbps, while HSDPA trials have shown promise at speeds of 400–700 kbps. T-Mobile is planning to deploy HSDPA through an upgrade of its GPRS networks and provide speeds of 384 kbps to 1.8 Mbps. On the other hand, the CDMA2000 carriers, i.e., Verizon Wireless, Sprint Nextel, and Alltel, have all moved to CDMA2000 EV-DO services. These services are characterized with speeds of 400–700 kbps with bursts of up to 2Mbps. Verizon has launched its streaming service V-Cast, which is based on Windows Media technology.
6.4 STANDARDIZATION FOR CARRIAGE OF MULTIMEDIA OVER 3G NETWORKS
We now take a look at the standardization for the call set up, multimedia data transfer, and video streaming or video multicasting, which makes mobile TV and multimedia services possible. These standards have been formalized under the 3G partnership fora.
With the two types of platforms and evolution paths for 3G, the standardization efforts have also progressed under two streams in the 3G partnership projects. The 3GPP has been closely associated with the developments of GSM networks to UMTS. The migration of CDMA-based networks to 3G is being coordinated by the 3GPP2.
There are significant differences between the two, i.e., 3GPP and 3GPP2, though unification standards are being strongly advocated for interoperability and interworking of applications. The process involved in the migration of networks from 2G, which are essentially circuit-switched voice-based networks with SS7 signaling, toward 3G networks with an IP core implies major upgrades to the network. The objective is achieved in stages. The 2.5G networks (GPRS and EDGE) in the GSM family and CDMA2000 1× and 3× had added packet-switching capabilities as an overlay without other structural changes such as voice coding standards. Unfortunately the hybrid networks so created still left the networks wanting for handling higher throughput data rates. The evolved 3G platforms have corrected these inefficiencies by moving to an IP core for both voice and data, using a more efficient coding for voice and migrating all signaling protocols to IP-based technologies.
The efforts have resulted in the following IP-based multimedia platform architectures for the networks:
• IP Multimedia System (IMS) 3GPP, and
• Multi-Media Domain 3GPP2.
Both these architectures provide for the transmission of voice and video over IP with quality of service (QoS) guarantees (Fig. 6-6).
FIGURE 6-6 Standardization of 3G Services
There has been an active effort toward the convergence of the two standards of 3GPP and 3GPP2, particularly with a view toward inter-working and roaming, some of the attributes that were lacking in the different technologies of the 2G systems involving GSM, TDMA, CDMA, and iDEN. The mobile receivers designed for interoperability will need to be multimode as well as multiband to be able to cater to the two interfaces (WCDMA and CDMA2000). However, the convergence is far from complete, with the need to support the technologies of WCDMA frequency division duplex (FDD) and TDD, which have different propagation characteristics and consequently network designs.
6.4.1 3GPP Standards
The 3GPP specifications for UMTS were initially finalized in 1999 and primarily addressed the new radio interface WCDMA and new radio access network architecture. It laid down the specifications of the audio and video codecs, media file format .3gpp, and session establishment procedure. The formalization of the 3GPP release ’99 laid the framework for the networks being able to establish video-calling or video-streaming sessions using standardized video and audio codecs and the common media file format .3gpp. Video streaming could be set up using a common specification called 3GPP-PSS. Release 4 in 2001 made further improvements in radio interfaces, UMTS transport, and architecture. However, the major evolution came with release 5, which included support for HSDPA and introduced the IP Multimedia System. It also introduced the IP Multimedia System (IMS) and the IP UMTS terrestrial radio access network (UTRAN) concept. The IP UTRAN uses the IP as a transport protocol for wireless traffic and realizes greater network efficiencies as well as routing flexibilities. Release 6 was a major enhancement again, with support for H.264 and the MBMS.
6.4.2 The IP Multimedia System
The IMS provides for an IP core network and packet-switched protocols for signaling (session initiation protocol or SIP). Owing to major changes in architecture the cost of migration of 3G remains high.
The IP Multimedia System is designed as a service network for handling signaling. The actual traffic takes a separate route. The IMS also mandates the use of IPv6, which adds significant capabilities for handling of multimedia traffic.
The handling of signaling information is based on the use of SIP and a proxy server. The sessions set up are independent of the content type that will be used. The IMS provides support for voice-over IP (VoIP) calls, which have already been commercially offered at a significant advantage.
6.5 MOBILE TV STREAMING USING 3GPP STANDARDS—PACKET-SWITCHED STREAMING SERVICE
Providing mobile TV or video via streaming is one of the most popular methods of video delivery. Streaming uses the packet-switching features of the underlying data networks.
Codec Formats for Packet-Switched Streaming Service Release 4
Mobile TV streaming via the 3G networks is governed by the 3GPP standards termed 3GPP-PSS. The basic purpose of the specifications for the streaming service is that there should be uniformity in:
• definition of video and audio formats to be handled,
• definition of coding standards,
• definition of call setup procedures for the streaming service,
• definition of protocols for the streaming service,
• QoS issues, and
• digital rights management for the streamed services.
The 3GPP-PSS services have evolved from a simple packet-streaming service as in release 4 of the 3GPP-PSS specifications (2001) to more advanced services while maintaining backward compatibility. The latest release is release 6 (2005).
Table 6-2 describes the specifications of audio and video formats and codecs for streaming service PSS release 4. As may be seen, video is prescribed to have a maximum frame size of 176 × 144 pixels and a coded bit rate of 64 kbps (maximum) (Fig. 6-7).
FIGURE 6-7 Mobile TV Streaming Architecture
6.5.1 Unicast Session Set Up in 3GPP Using Packet-Switched Streaming
The procedure for setting up unicast real-time streaming protocol (RTSP) sessions between a mobile device and a streaming server is shown below. The client on the mobile (e.g., HTTP client) selects the location of a media file with an RTSP URL. The following sequence of events takes place:
The media player connects to the streaming server and gives an RTSP describe command.
The server responds with a session description protocol message (SDP) giving the description of media types, number of streams, and required bandwidth.
The player or the media client analyzes the description and issues an RTSP SETUP command. This command is issued for each stream to be connected.
After the streams are set up the client issues a PLAY command. On receiving the PLAY command the streaming server starts sending the RTP packets to the client using UDP.
FIGURE 6-8 Streaming Session Setup in 3GPP-PSS
FIGURE 6-9 3GPP Packet Streaming Service Protocol Stack
The connection is cleared by the client when desired by issuing a TEARDOWN command.
The RTP, RTCP, RTSP, and SDP commands are as per the relevant RFC standards (Figs. 6-8 and 6-9).
There have been further enhancements to the PSS protocol and features in the new releases. Release 5 of the PSS has introduced the concept of the user agent profile. Using this feature the client on the mobile can signal to the server its capabilities in terms of the number of channels of audio, media types supported, bits per pixel, and screen size. Using this information the streaming server can connect the appropriate streams to the client. The above information is furnished during the session initiation. Release 5 has also added new media types, including synthetic audio (MIDI), subtitles (time-stamped text), and vector graphics.
6.5.2 Progressive Download
PSS release 6 (2004) added a number of new features for reliable streaming and, most importantly, digital rights management. New protocols have been proposed for reliable streaming, which include features for retransmission of information including progressive download using HTTP and RTP/RTSP over TCP. These ensure that no information is lost in the streaming process. This type of streaming is suitable for downloads and less so for live TV. The PSS protocols have also been enhanced to provide QoS feedback to the server. This conveys the information on lost packets, error rate, etc. New codec types, i.e., MPEG-4/AVC or H.264 and Windows Media 9, have also been recommended.
PSS release 6 also requires support of digital rights management as per 3GPP-TS 22.242.
6.6 UNIVERSAL MOBILE TELECOMMUNICATION SYSTEM
UMTS uses an air interface called the UMTS terrestrial radio access network, which has been standardized for WCDMA systems under the IMT2000 framework. The core network of the GSM/GPRS remains largely unchanged for backward compatibility.
UMTS networks also provide the feature of handovers between GSM/GPRS and UMTS networks, facilitating compatibility and roaming. UTRANs operate in the 1920- to 1980- and 2110- to 2170-MHz bands and can be added to the existing GSM base stations. Operation in other bands is also possible as described in Chap. 10. The GSM base station system and the UTRAN share the same GPRS core network comprising the SGSN and GGSN. The radio network controllers (RNCs) are connected to the mobile switching centers, and, in the case of UTRAN, to the SGSN for UMTS (3G) (Fig. 6-10).
FIGURE 6-10 UMTS/GSM Mobile Network Architecture
6.6.1 The UMTS Core Network
The WCDMA technology involved a major change in the radio interfaces from the previous GSM/GPRS networks. The wideband Direct Sequence Code Division Multiple Access uses a bandwidth of approximately 5 MHz. There are two basic modes of operation—in the FDD mode separate 5-MHz frequencies are used for the uplink and downlink. This mode thus uses the paired spectrum for UMTS. In the TDD mode the same 5-MHz bandwidth is shared between uplink and downlink and is primarily intended to use the unpaired spectrum for UMTS.
The frame length is 10 ms and each user is allowed one frame during which the bit rate is kept constant. However, the user can vary the bit rate from frame to frame, which lends the system the ability to provide bandwidth on-demand-based services. The WCDMA radio interface is grouped into UTRAN, which was a major change from the older GSM/GPRS networks, though the handover between the UTRAN and the GSM networks was provided for. However, in the first release of the UMTS core network architecture (release ’99) the architecture was largely inherited from the GSM core networks.
6.6.2 UMTS Release ’99 Core Architecture
Release ’99 core architecture has a provision for two domains, i.e., the circuit-switched (CS) domain and packet-switched (PS) domain. The CS domain supports the PSTN/ISDN architectures and interfaces, while the PS domain connects to the IP networks.
6.6.3 Video Coding Requirements for Transmission on 3G Networks
Mobile networks present a challenging environment for the transmission of video to individual users in a 3G (or cellular network). The carriage of video over 3G networks falls into the following categories:
1. Video in multimedia message services (MMS): Messages are sent from the server by streaming to the mobile receiver.
2. Video in live or streaming mode: Video is transferred in a unidirectional mode (half-duplex).
3. Video in conversational services, including video conferencing: This requires the transfer of video and audio in full-duplex mode.
The carriage of video on 3G requires a channel on which the bit rate can be sustained to meet the minimum video carriage requirements. In 3G networks there is only limited capacity to support streaming video as it is very bandwidth intensive.
6.6.4 UMTS Quality of Service Classes
UMTS networks have the following classes of traffic, which are distinguished primarily by how sensitive they are to the delays that might be experienced in the network:
• Conversational class,
• Streaming class,
• Interactive class, and
• Background class.
The Conversational class is the most delay sensitive, while the Background class is used for non-real-time services such as messaging.
The Conversational class is designed for speech and face-to-face communications such as video telephony, for which the acceptable delay should not exceed 400 ms. In UMTS, the Conversational class is used to provide the AMR speech service as well as the video telephony service (H.324 (mobile) or 3G-M324). The speech codecs in UMTS use the adaptive multirate coding technique (AMR).
The AMR codec can be controlled by the radio access network to enable interoperability with the existing 2G cellular networks such as GSM (EFR codec 12.2 kbps) or the U.S. TDMA speech codec (7.4 kbps). The bit rates possible for AMR are 12.2, 7.95, 7.4, 6.7, 5.9, 5.15, and 4.75 kbps.
Video telephony standards for PSTN networks are prescribed by ITU-T H.324 recommendations and have been in use in video telephony and conferencing applications for a long time. These are used over PSTN/ISDN connections. The H.324 uses H.263 as video codec, and G.723.1 (ADPCM) as speech codec. The audio, video, and user data is multiplexed using an H.223 multiplexer, which gives a circuit-switched bit rate of n × 64 kbps.
Conversational calls on mobile 3G networks have followed a modified version of the PSTN standard called H.324(M) or 3G-324M. The standard has agreement from both 3GPP and 3GPP2 fora and is in use on 3G networks for the Conversational class.
6.6.5 3G-324M-Enabled Networks
While the objectives of the 3GPP and 3GPP2 projects are to move toward an IMS (3GPP) and multimedia domain system (3GPP2), both based on IP core, for the initial implementation of the carriage of video over 3G networks, the initial agreement and convergence between the two groups have been on the use of an agreed-upon standard, 3G-324M, for the transport of video over 3G networks. Both the 3GPP and the 3GPP2 organizations have adapted the 3G-324M protocol as the means of transporting conversational video over mobile networks, e.g., 3G. 3G-324M (also known as H.324 annex C) envisages initial mobile services using 3G data bandwidth without IP infrastructure. The service uses the 64 kbps data channel, which provides an error-protected and constant bit rate interface to the application (Fig. 6-11).
In the mobile network, 3G-324M-based video content is carried by a single H.223 64-kbps stream that multiplexes audio, video, data, and control information. In accordance with 324M, the video portion of the H.223 protocol is based on MPEG-4, whereas the audio portion is based on NB-AMR coding. The control portion of the H.223 stream is based on the H.245 protocol, which is primarily responsible for channel parameter exchange and session control.
The constant bit rate interface simplifies the interface as the bit rates achieved are not dependent on the number of active users in the network for data services.
3G services in Japan were launched using the 324M standard, i.e., without the IP multimedia system.
6.6.6 UMTS Streaming Class
Streaming class is very popular in Internet applications as it allows the receiving client to start playing the files without having to download the entire content. The receiver (e.g., a media player) maintains a small buffer, which helps a continuous playout of content despite transmission packet delays and jitter. The streaming applications fall into two broad categories: Web broadcast (or multicast) and unicast. As the name suggests, the Web casting is for an unlimited number of receivers, all of which receive the same content, while unicast is a server-to-client connection by which the client can communicate with and control the packet rates, etc., and request retransmission of missed packets (Fig. 6-12).
FIGURE 6-13 Multicast Mobile TV
Multicasting in cellular networks implies constant usage of a certain bandwidth in every cell, which reduces the capacity available for other purposes and uses by an equivalent amount (Fig. 6-13).
The 3G streaming TV services for live TV (such as MobiTV) are subject to handset as well as network limitations. The resources are very limited, even in 3G networks, and the initial offering, e.g., by Sprint Nextel, offered a frame rate of 7 fps. The rate dropped lower to 1–3 fps for some networks using Java applets for delivery.
Streaming video is delivered via the RTSP protocol, which allows the multimedia streams delivered to be controlled via RTP (Figs. 6-14).
6.6.7 Interactive Class
The Interactive class is designed for user equipment applications interacting with a central server, another remote application. Examples of such applications are Web browsing or database access. Round-trip delay in the Interactive class of applications is important but not so critical as in the Conversational class.
6.6.8 Background Class
The Background class is meant for applications that are not delay sensitive such as e-mail, SMS, and MMS services. Of particular interest in UMTS are the MMS messages, which can carry multiple elements as content including text, images in any format(GIF, JPEG, etc.), audio and video clips, and ring tones. MMS messages are carried as per the 3GPP and the WAP forum standards.
FIGURE 6-14 Wireless Streaming Architecture
6.7 DATA RATE CAPABILITIES OF WCDMA NETWORKS
WCDMA is one of the two main technologies for the implementation of third generation (3G) cellular systems, the other being CDMA2000 and its evolutions. The UMTS frequency bands assigned for WCDMA are 1920–1980 and 2110–2170 MHz (frequency division duplex), with each frequency being used in a paired mode with 5 MHz (i.e., 2 × 5 MHz). This can accommodate approximately 196 voice channels with AMR coding of 7.95 kHz. Alternatively, it can support a total physical level data rate of 5.76 Mbps. In various networks the users can be offered data rates from 384 kbps to 2.4 Mbps (spreading factor 4, parallel codes (3 DL/6 UL), 1/2 rate coding), depending on the usage patterns, location of the user, etc. There are higher rate implementations, i.e., HSDPA, with data rates of 10 Mbps or higher, as described later, as well as 20 Mbps for MIMO systems.
Channel bandwidth |
5MHz |
Chip rate |
3.84Mcps |
Frame length |
10msec |
No. of slots per frame |
15 |
No. of chips per slot |
2650 |
RF structure (forward channel) |
Direct spread |
The data transmitted on WCDMA systems is in the form of frames with each frame being of 10 msec. The channel data rate is 5.76Mbps and with the QPSK coding used this gives a chip rate of 3.84 Mchips/sec or a frame chip capacity of 38,400 chips. Each frame has 15 slots, which can carry 2560 chips (Table 6-3).
6.7.1 Data Channels in UMTS WCDMA
In the uplink direction the physical layer control information is carried by a dedicated physical control channel, which has a spreading factor of 256 (i.e., bit rate of 15 kbps) in 5-MHz WCDMA systems. The user data and higher layer control information are carried by dedicated physical data channels, which can number more than 1. These physical data channels can have different spreading factors ranging from 4 to 256 depending on the data rate requirements.
The transport channels, which are derived from the physical channels, can be divided into three categories:
• common channels,
• dedicated transport channels, and
• shared transport channels.
The channels are described below.
Common channels: In the uplink the common access channel is the random access channel (RACH), while in the downlink it is the forward access channel (FACH). Both channels carry signaling information and in addition can also carry data. Typically these channels have a short setup time, but do not offer any closed loop power control. These are suitable for sending a few IP packets. There are typically one or more channels of each type in each cell area.
Dedicated channel (DCH) uplink and downlink: These channels require a setup procedure and hence there is a small delay in their setup. They can be used for any data rate up to 2 Mbps based on the resources available. Closed-loop power control is used in these channels and hence their use is efficient and less prone to generating interference in adjacent cell areas. The dedicated channels can have a variable bit rate on a frame-to-frame basis.
Shared channel (DSCH) downlink: This channel is time division multiplexed with quick setup and fast power control. It is suitable for large and “bursty” data up to 2 Mbps. The data rate can be varied on a frame-to-frame basis.
The DCHs are not very well suited to very bursty IP data as the setup time of any reconfiguration changes in the channel can be as much as 500 ms (Fig. 6-15).
6.7.2 Classes of Service in WCDMA 3GPP
Table 6-4 lists the classes of service that are supported in the 3GPP release ’99 architecture of UMTS under 3GPP standardization.
6.7.3 Release 5 Core Network Architecture and IP Multimedia System
Release 5 of the 3GPP core network has major enhancements in the support of packet- and IP-based architecture. It also supports phase 1 of the IP multimedia system. The major enhancement in release 5 of the 3GPP was the introduction of the HSDPA technology to increase the packet data handling rates. The primary mechanism of the data rate increase is the fast physical layer transmission using 8PSK modulation as well as link layer enhancements using fast link adaptation (Fig. 6-16).
FIGURE 6-15 WCDMA Transport Channels
Classes of Service in WCDMA
Sprint Nextel was the first to launch 3G services based on HSDPA. Cingular Wireless followed and already had launched a service based on HSDPA in December of 2005 and it is now progressively being ramped up to cover its entire network.
FIGURE 6-16 Release 5 of UMTS Core Network Architecture
6.7.4 3GPP Release 6
The 3GPP issued release 6 in June 2005 as part of the continued evolution of the 3G UMTS networks. 3GPP release 6 includes the MBMS. Using MBMS the content is broadcast using IP data casting so that a large number of users can receive the services without tying up the network resource in unicast connections.
HSDPA is a feature added in release 5 of the 3GPP specifications. HSDPA extends the DSCH, allowing packets destined for many users to be shared on one, higher bandwidth channel called the high-speed DSCH. To achieve higher raw data rates, HSDPA uses, at the physical layer, higher level modulation schemes such as 16-point quadrature amplitude modulation (16QAM), together with an adaptive coding scheme. The HSDPA also changes the control of the medium access control (MAC) function from the radio network controller to the base station. This allows the use of fast adaptation algorithms to improve channel quality and throughput under poor reception conditions. On the average download speeds for DSCH can be 10 Mbps (total shared among the users). However, lab tests and theoretical predictions suggest the rates can be as high as 14.4 Mbps. Of course the maximum data rate falls as the users move outward in the cell and can fall to 1–1.5 Mbps at the cell edge. HSDPA also uses IPv6 in the core network, together with improved protocol support for bursty traffic.
6.8.1 Data Capabilities of the HSDPA Network for Video Streaming
Under normal conditions the HSDPA network can deliver 384 kbps to up to 50 users in a cell area, which is a 10-fold improvement over the release ’99 WCDMA, with which only 5 users could be provided such throughput (Fig. 6-17).
As per an analysis (Ericsson) of HSDPA networks with 95% of satisfied users, 128 kbps streaming service can be provided at 12 erlangs of traffic. Under low usage conditions (i.e., 2 × 5 min per day) all the users in the cell area (assumed user density per cell of 600) can get satisfactory service. For medium usage (assumed 5 × 10 min per day) the users that can be catered to within the satisfaction level falls to 171 per cell or 28%, while for high usage (e.g., 4 × 20 min) the usage falls to 108 users per cell or 18%. The unicast services do have the advantage that the number of content channels can be virtually unlimited (including the video on-demand channels) as no resource is used in the idle condition. When a user sets up a connection with the server the content is delivered and resources are used. However, as the figures above indicate, the unicast service does not scale well with the number of users or high-usage patterns. Other disadvantages (advantage for network operators) is that the users incur the data transmission charges.
FIGURE 6-17 Capability of HSDPA Network to Cater to Simultaneous Users
6.8.2 System Capability of 3G WCDMA Network for Video Streaming
The data handling capability of 3G systems for streaming data is limited for applications for which a minimum data rate must be maintained.
6.9 MULTIMEDIA BROADCAST AND MULTICAST SERVICE
The potential limitations of the 3G networks for streaming of high-usage TV traffic has led to the consideration of multicast technologies that are inherently more suitable and less resource intensive, particularly for live TV channels. The live TV channel traffic is essentially of multicast nature, with all users viewing identical streamed content. Multicast networks are ideally suited to such delivery. In multicast networks, each content channel is allocated one transport channel in each cell area irrespective of the number of users watching.
In an MBMS service all routers need to repeat the multicast transmission in each cell. It is estimated that one 64-kbps multicast channel requires approximately 5% of the carrier power, while a 128K channel requires 10% of the carrier power. This implies that up to 10 × 128K or 20 × 64K multicast channels can be supported per carrier in a cell area (the number of channels can vary depending on cell topography and type of receivers used). MBMS is an in-band broadcast technique as opposed to other broadcast technologies for mobile TV such as DVB-H. MBMS uses the existing spectrum of the 3G by allocating spectrum or carrier resources to the multicast transport channels in each cell.
The cells can have both unicast and multicast channels, depending on the traffic dimensioning. The MBMS is essentially a software-controlled feature that enables the dedication of transport channels to multicast TV. Typically only those channels that are of high viewership interest would be multicast in MBMS networks (Fig. 6-18).
FIGURE 6-18 Audiences for Broadcast and Unicast Channels
As the name suggests, the multimedia broadcast and multicast service operates in two modes: Broadcast mode is available to all users without any differentiation (such as payment status). The users can receive the channel with a requested QoS. In the multicast mode the channel is available to select users in a selected area only. The users can receive the service based on payments or subscription.
The MBMS service setup can be explained by the following setup procedure:
• Service announcement: Operators would announce the service using advertising or messaging, etc. The announcement may go only to subscription customers in the case of multicast services.
• Joining: The multicast users can indicate that they would be joining the service. The joining can be at any time but only the authorized users will receive the multicast service. In broadcast mode all users would receive the service.
• Session starts: The requisite resources are reserved in the core network as well as the radio networks.
• MBMS notification: A notification goes out of the forthcoming service.
• Data transfer: The data transfer commences and is received by all users in the selected group. In broadcast mode all users would receive the data, which is without any encryption. In multicast mode the data is encrypted and only the authorized users receive the service.
• Leaving or session end: In multicast mode the users may leave the session at any time, or the session ends after the data transfer is completed (Fig. 6-19).
FIGURE 6-19 MBMS Delivery—MBMS Reserves Transport Capacity in All Cells for Multicast TV
6.10 MOBILE TV SERVICES BASED ON CDMA NETWORKS
The UMTS (WCDMA) is the evolution path on which the 3G-GSM-based operators have moved ahead to launch the mobile TV across the globe. The evolution path for the CDMA (IS-95)-based networks has been to move toward higher data rates through the CDMA2000 framework with multicarrier options. The CDMA2000 1×–CDMA2000 3×–CDMA1×-EV-DO technologies reflect this growth curve.
The standards for 3G networks that have evolved from CDMA networks are being developed by the 3GPP2 forum. These are also being simultaneously developed under the ITU-R IMT2000 CDMA framework. The multicarrier approach was adapted for compatibility with the existing IS-95 networks with 1.25 MHz carrier spacing. This was particularly the case for the United States, where no separate spectrum for 3G UMTS services was available.
In multicarrier mode transmissions, in the downward transmission direction, up to 12 carriers can be transmitted from the same base station, with each carrier being of 1.25 MHz and a chip rate of 1.2288 Mcps. The 3 × version with three carriers can therefore provide a data rate of 3.6864 Mcps, which compares well with the UMTS (WCDMA) chip rate of 3.84 Mcps. The CDMA2000 standards of the ITU have adapted data rates for up to 3 carriers, with the CDMA2000 3 × standards having been defined in the CDMA2000 standard. The multiple carrier approach provides compatibility with the IS-95 networks, which have the same carrier bandwidth and chip rate.
Another branch of developments has been the 1× mode with the evolution of the 1×EV-DO (data-only option). This system uses a separate carrier for voice and data services. The peak data rate in the EV-DO carrier is 3 Mbps using a bandwidth of 1.25 MHz (Fig. 6-20).
The 1×EV-DO networks achieve high throughputs with a 1.25-MHz bandwidth by using advanced modulation and RF technology. This includes first an adaptive modulation system that allows the radio node to increase its transmission rate based on the feedback from the mobile. It also uses advanced “turbo-coding” and multilevel modulation, which acts to increase the data rates at the physical layer. It also uses macrodiversity via a sector selection process and a feature called multiuser diversity. The multiuser diversity permits a more efficient sharing of resources among the active users.
The core architectures of the 1× EV-DO networks are also moving toward an IP core with a packet-routed network for efficient handling of data. The 1×EV-DO networks can provide average user data rates for downloads at 300–600 kbps, with peak data rate capability of the network being 2.4 Mbps. EV-DO uses multilevel adaptive modulation (modulation used is QPSK, 8PSK, and 16QAM.).
FIGURE 6-20 Evolution of IS-95 Services to 3G
6.10.1 1×EV-DO Architecture
Figure 6-21 gives the 1×EV-DO architecture in which the radio nodes installed at cell sites perform packet scheduling, base-band modulation/demodulation, and RF processing. Handoff functions are provided by the radio network controller installed at the central office. The radio network controllers connect to the core network using a packet data serving node (PSDN).
The 1×EV-DO network goes beyond the circuit-switched architecture of IS-95 networks (a feature that still exists in 3G release ’99 architecture). This flexibility has led the operators to build their networks based on an IP core for switching, transport, and backhaul applications. The use of an IP core leads to better cost efficiency through the use of standard routers and switches rather than proprietary equipment.
FIGURE 6-21 1×EV-DO Network Architecture
6.10.2 Mobile TV Using 1×EV-DO Technologies
1×EV-DO is already an international standard under the 3GPP2 forum and has been deployed in a number of networks in Japan (KDDI), Korea (SK Telecom and KTF), and the United States (Verizon and Sprint) as well as other countries. 1×EV-DO does not provide compatibility to CDMA2000 networks. This needs to be achieved using multi-mode handsets. The ready availability of handsets having compatibility with CDMA2000 networks has led to an increasing use of the services of 1×EV-DO. The users have the flexibility to receive incoming voice calls (CDMA2000 1×) while downloading data using 1×EV-DO. Newer versions of handsets also provide support for GSM/GPRS networks. The enhancement of CDMA2000 to 1×EV-DO is very easy by addition of a channel card to the existing base stations.
The 1×EV-DO networks have the flexibility to support both user- and application-level QoS. Applications such as VoIP can be allocated priority using application-level quality of service. This helps this delay-sensitive service to work well even in high-usage environments. User-level QoS allows the operator to offer premium services such as mobile TV. The 1×EV-DO packet scheduling combined with Diff-Serv-based QoS mechanisms can enable QoS within the entire wireless network.
Verizon has launched its VCAST service, which provides streaming audio and video clips (news, weather, entertainment. and sports). The service using 1×EV-DO can be delivered at 400–700 kbps with burst speeds of 2 Mbps. Verizon uses Windows Media 9 as the delivery mode.
6.10.3 CDMA 1×EV-DV Technology
Mention must be made of the 1×EV-DV (data and voice) networks as these provide compatibility with the CDMA2000 architecture. The operation can be extended to 3× mode with multicarrier operation, and data rates (peak) of 3.072 Mbps downlink and 451 kbps uplink can be achieved. The system has advanced features such as adaptive modulation and coding (QPAK, 8PSK, and 16QAM) and variable frame duration. The mobile device can select any of the base stations in its range.
6.10.4 CDMA2000 1×EV-DO Networks
1×EV-DO has already been deployed widely across the United States (Verizon, Sprint, and Alltel), Canada (Bell Mobility), Korea (SK Telecom and KT Freetel), Japan (KDDI), Australia (Telstra), New Zealand (Telecom New Zealand), and a number of other countries. Sprint and KDDI have already moved to 1×EV-DO Rev A and others are in the process of launching Rev A services. These operators are now leveraging the data capabilities of the networks to deliver a number of innovative multimedia services such as music/video streaming, videophone, and live TV broadcast.
In the United States Sprint and Verizon are the largest operators, with coverage of over 200 cities each.
6.11 Wi-Fi MOBILE TV DELIVERY EXTENSIONS
Recently there has been growing interest in wireless LAN technologies, especially in 802.11, also known as Wi-Fi, and the 802.16 (WiMAX). Low cost of equipment and wide availability of subscriber devices in laptops has led to a proliferation of Wi-Fi in homes for home networking and in the enterprise for in-building mobility. Hundreds of thousands of public hot spots have been rolled out using Wi-Fi technologies.
While it is capable of supporting high-speed Internet access with full mobility at pedestrian or vehicle speeds, 1×EV-DO is equally powerful for serving hot spots such as hotels, airports, and coffee shops. Therefore, with growing adoption of 1×EV-DO, there is no longer a need to use Wi-Fi hot spots. However, Wi-Fi and 1×EV-DO may continue to play complementary roles, as Wi-Fi serves the enterprise and home-networking markets, while 1×EV-DO provides the wide-area mobile data coverage.
Although the current implementations of 1×EV-DO are geared toward high-speed wireless data applications, with new enhancements being added, it is on its way to becoming an all-purpose IP-based air interface that can efficiently support any kind of IP traffic, including delay-sensitive multimedia traffic such as VoIP.
Other enhancements are also being developed for 1×EV-DO, including a much faster reverse link, with peak user rates in excess of 1 Mbps, and a broadcast capability to support applications such as news/music/video distribution and advertising. Advanced chip sets that support 1×EV-DO will also be able to support 1× voice and 1×EV-DO simultaneously, to allow a user to maintain a phone conversation while accessing the Internet at broadband speeds.
6.12 BROADCASTING TO 3GPP NETWORKS
The file formats used in multimedia constitute a wide range from raw audio and video files to MPEG files or Windows Media files (*.mpg, *.mpeg, *.mpa, *.dat, *.vob, etc.). On the other hand 3GPP networks are characterized by handling video, audio, and rich media data as per the file formats in GPP releases (releases 4, 5, and 6). The current 3GPP version, as of this writing, is release 6. The resolutions of QCIF and QVGA for video have been formalized under these standards and .3gp is the file format used in mobile phones to store media (audio and video). Audio is stored in AMR-NB or AAC-LC formats. This file format is a simpler version of ISO 14496-1 (MPEG-4) Media Format. The .3gp format stores video as MPEG-4 or H.263.
3GPP also describes image sizes and bandwidth, so content is correctly sized for mobile display screens. Standard software such as Apple’s QuickTime Broadcaster supports video codecs for MPEG-4 or H.264 and together with a QuickTime streaming server provides for a broadcasting solution that can reach any MPEG-4 player. A number of other broadcast solutions provide real-time MPEG-4 or 3GPP encoding and streaming applications as well as downloadable players.
6.12.1 QuickTime Broadcaster
Apple’s QuickTime Broadcaster comprises a Mac OS Xserver 10.4 with QuickTime 7 software and QuickTime Broadcaster. The QuickTime streaming server forms part of the software. The server supports MPEG-4, H.264, AAC, MP3, and 3GPP as permissible file formats. The server can serve streams with 3GPP on either a unicast or a multicast basis for mobile networks. Both modes of delivery, i.e., streaming or progressive download, are supported. The server uses industry standard protocols for streaming, i.e., RTP/RTSP. Live programs can be encoded using QuickTime Broadcaster. It is also possible to have on-demand streaming.
6.12.2 Model 4Caster Mobile Encoder Solution from Envivio
The Model 4Caster has been designed for broadcasting on mobile TV networks with multistandard support. The 4Caster can accept a video input in any format. It presents its output in eight simultaneous profiles, which include 2.5G, 3G, 3.5G, DVB-H, DMB, ISDB-T, and Wi-Fi or WiMAX networks. It supports both 3GPP and 3GPP2 network standards. The encoder outputs offer simultaneous streaming and broadcasting of mobile video at multiple bit rates. It offers live bit rate switching, which enables the encoder to adjust the bit rates based on network conditions. The encoder also supports content protection using ISMAcrypt.
FIGURE 6-22 A Typical 3GPP Headend
6.13 A TYPICAL 3GPP HEADEND FOR MOBILE TV
A typical 3GPP headend would comprise two servers:
• a broadcast server for encoding of audio and video content and IP encapsulation into IP UDP RTP packets and
• a streaming server for providing multiple unicast RTP streams to multiple mobile handsets (Fig. 6-22).
In a headend the video and audio sources can be a video cameras (e.g., a USB camera) or a satellite decoder/receiver or even stored video content or video tape. The broadcasting server encodes the stream using H.263 encoders and AMR audio coders and provides IP packets that travel via RTP/UDP. The video resolution of the encoded stream will be limited to 3GPP, e.g., a QCIF size with 15–30 fps. The audio encoding may be AMR 4.7 to 12.3 kbps.
The streaming server sets up one-to-one unicast connections to mobile sets whose users desire the particular video to be accessed by streaming. For this purpose the mobiles would access the Web site through a command such as rtsp://<server>/<filename>.
The audio video data is decoded at the receiving end (i.e., a mobile phone) using a 3GPP player embedded in the handset.