CHAPTER 2

RESOURCE ALLOCATION APPROACHES IN MULTI-TIER NETWORKS

2.1 INTRODUCTION

In a two-tier macrocell–femtocell network, femtocells are deployed over the existing macrocell network and share the same frequency spectrum with macrocells. Due to spectral scarcity, the femtocells and macrocells have to reuse the total allocated frequency band partially or totally which leads to cross-tier or co-channel interference. At the same time, in order to guarantee the required QoS to the macro users, femtocells should occupy as little bandwidth as possible that leads to co-tier interference. As a result, the throughput of the network will decrease substantially due to such co-tier interference and CTI. In addition, severe interference may lead to “deadzones,” that is, areas where the QoS degrades significantly. The deadzones are created due to asymmetric levels of transmission power within the network and the distance between macrocell UE and macrocell base station. For example, a macrocell UE located at a cell edge and transmitting at a high power will create a deadzone to the nearby femtocell UL transmission due to co-channel interference. On the other hand, in the DL transmission, due to high path-loss and shadowing, a cell-edge macrocell UE may experience severe co-channel interference from the nearby femtocells. Thus, it is essential to adopt an effective and robust interference management scheme that would mitigate the co-tier interference and reduce the CTI substantially in order to enhance the throughput of the overall network. In this chapter, we will review the existing literature on resource allocation and interference management in two-tier macrocell–femtocell networks.

In OFDMA-based small cell networks, due to the flexibility in spectrum allocation, orthogonal sub-carriers can be assigned to small cells and macrocells. This gives OFDMA-based small cells an edge over CDMA-based small cells in terms of utilizing the spectrum resources efficiently. In OFDMA-based wireless networks, radio resource allocation (e.g., sub-channel and power allocation), CAC, and power control algorithms are crucial to optimize the resource utilization while providing support to a wide range of multimedia broadband services with heterogeneous QoS requirements. In general, the radio resource management (RRM) strategies for OFDMA-based networks can be grouped into three categories [1]. The first category includes frequency/time resource allocation schemes such as channel allocation, scheduling, transmission rate control, and bandwidth reservation schemes. The second category consists of power allocation and control schemes, which control the transmission power of the UEs and the BSs. Note that the channel and power allocation schemes (i.e., the resource allocation schemes) can be referred to as interference management schemes. The third category comprises of CAC, BS assignment, and handoff algorithms, which control the accessibility of the UEs to the radio network. The CAC is responsible for admission or rejection of incoming request to the network based on predefined criterion while taking the network load conditions and QoS requirements of both incoming and ongoing users into account. Then, scheduling techniques assign available resources to the admitted users. CAC takes into account the amount of resource available in the cell, QoS requirements of the new users as well as their priority levels, and the currently provided QoS to the active sessions in the cell. A new request is only granted resources if it is estimated that the QoS requirement for the new request can be fulfilled, while still being able to provide acceptable QoS to the existing in-progress sessions in the cell. While we focus on the channel allocation and interference management issues in this chapter, we will deal with the CAC issue in multi-tier cellular networks in Chapter 6.

2.2 DESIGN ISSUES FOR RESOURCE ALLOCATION IN MULTI-TIER NETWORKS

In this section, we highlight the major performance measures and issues which are considered when designing radio resource allocation in traditional OFDMA-based networks in general. The same issues arise in the context of multi-tier OFDMA-based networks.

Minimum rate outage: When the transmission rate of a selected user is not satisfied, a rate failure (or rate outage) occurs. The minimum rate required for each user has to be achieved with a predefined rate outage.

Channel utilization: Channel utilization, a term related to the channel usage, includes not only the data bits but also the overhead that makes use of the channel. A resource allocation method should minimize the overhead to maximize the channel utilization.

Bit rate-related parameters: The maximum sustained and the minimum reserved traffic rates (measured in bits per second) are bit rate related QoS parameters. The former is the peak information rate of the service flow whereas the latter is the minimum information rate promised to the service flow.

Fairness: It is one of the challenging problems for RRM. In general, there is trade-off between channel utilization and fairness. Increasing the channel utilization, for example, by serving users with good channel conditions, results in unfairness.

Delay-related parameters: These include maximum latency and tolerated jitter. The maximum latency measures the maximum time elapsed or delay between when a sending data unit enters the medium access control (MAC) layer and when it enters the air interface. On other hand, the tolerated jitter determines the maximum variation in delay.

Call-level and packet-level QoS parameters: Both call-level and packet-level QoS parameters need to be considered for resource allocation. Call dropping probability and handoff call dropping probability are considered as call-level QoS metrics while packet throughput, packet delay, packet dropping rate, and delay jitter are considered as packet-level QoS metrics.

Policy-related parameters: Both traffic priority and request/transmission policy relate to policy-based parameters. Traffic priority is the priority given to a service flow, for instance, if two services have identical QoS parameters, preference is given to the service flow that has higher priority. The request/transmission policy parameters specify the service flow attributes.

Pre-emption strategy: This strategy is used to pre-empt the assigned bandwidth to admitted users with low priority services for new high priority service users. Cut-off and service degradation are two basic schemes used for pre-emption. In the former case, when pre-emption is executed, the total bandwidth assigned to low priority traffic is abruptly pre-empted and as a result its QoS can no longer be sustained. In the latter scheme, to pre-empt the bandwidth that has been assigned to admitted low priority service users, bandwidth degradation is performed with some predefined degradation level.

Differentiated services: One of the most challenging issues for wireless networks is to support multiple classes of services with different QoS requirements. Efficient resource management techniques are the keys to support multimedia applications with QoS provisioning.

Adaptive bandwidth/power allocation: Adaptive bandwidth/power allocation is essential due to time-varying nature of wireless channel, diversity of applications and QoS requirements to improve utilization of wireless network resources.

Energy efficiency (EE): EE is the overall number of bits transmitted per unit energy and is equivalent to the sum capacity of users per unit power. To calculate EE, in addition to the energy consumption for transmission and reception, the circuit energy consumption should also be taken into account. It depends on network architectures, transmission schemes, and resource allocation strategies.

2.3 INTERFERENCE MANAGEMENT APPROACHES

Figure 2.1 illustrates all possible interference scenarios in an OFDMA-based two-tier macrocell–femtocell network. In this network, each UE communicates on a specific sub-channel with the BS from which it receives the strongest signal strength, while the signals received from other BSs on the same sub-channel are considered as interference. Two types of interferences occur in such a multi-tier network.

Co-tier interference: This type of interference occurs between neighboring femtocells. For example, a femtocell UE (aggressor) causes UL co-tier interference to the neighboring femtocell base stations (victims) (e.g., index 5 in Figure 2.1). On the other hand, a femtocell base station acts as a source of DL co-tier interference to the neighboring femtocell UEs (e.g., index 6 in Figure 2.1).

Cross-tier interference (CTI): This type of interference occurs between femtocells and macrocells. For example, femtocell UEs (referred to as FUEs) and macrocell UEs (referred to as MUEs) act as sources of UL CTI to the serving MeNB (e.g., index 3 in Figure 2.1) and the nearby HeNBs (e.g., index 1 in Figure 2.1), respectively. On the other hand, the serving MeNB and HeNBs cause DL CTI to the FUEs (e.g., index 2 in Figure 2.1) and nearby MUEs1 (e.g., index 4 in Figure 2.1), respectively.

If an effective interference management scheme can be adopted, then the co-tier interference can be mitigated and the CTI can be reduced which would enhance the throughput of the overall network.

Different techniques such as cooperation among macrocell BSs (i.e., MeNBs) and femtocell BSs (i.e., HeNBs), formation of groups of HeNBs and exchange of information (such as path-loss, geographical location, etc.) among neighboring HeNBs, accessing the spectrum intelligently, etc. can be considered to reduce co-tier interference and CTI. In the following, we provide an overview of the different approaches for interference mitigation in two-tier OFDMA networks. These approaches consider UL and/or DL transmissions as well as co-tier interference and/or CTI. In the following sections, we give an overview of the different interference management techniques for two-tier macrocell–femtocell networks presented in the recent literature. To this end, we will provide a qualitative comparison among these techniques based on some important criteria. We will conclude the chapter and outline some open challenges related to interference management in multi-tier networks.

2.3.1 Femto-Aware Spectrum Arrangement Scheme

In [2], Yi Wu et al. propose a femto-aware spectrum arrangement scheme to avoid UL CTI between a macrocell and femtocells. In this scheme, the allocated frequency spectrum for any macrocell coverage area is divided into two parts: the macrocell dedicated spectrum and macrocell–femtocell shared spectrum. It is assumed that shared spectrum allocated to femtocells (i.e., HeNBs) is configured by the mobile operator. Thus, the macrocell base station (i.e., MeNB) has adequate knowledge of the shared frequency spectrum. Based on this knowledge, the MeNB develops an interference pool which includes the macrocell UEs that pose a threat to the nearby HeNBs. These macrocell UEs are thus assigned a portion of the spectrum dedicated for macrocell usage which reduces/mitigates the UL CTI and solves the UL deadzone problem.

Figure 2.2 illustrates the femto-aware spectrum arrangement scheme, where macrocell UE4, macrocell UE5, and macrocell UE6 pose potential threat of CTI on their prospective nearby HeNBs. Therefore, these macrocell UEs are put into the femtocell-interference pool by the MeNB and are assigned a dedicated portion of the total frequency spectrum in order to mitigate co-channel interference. On the other hand, since other macrocell UEs (i.e., macrocell UE1, macrocell UE2, and macrocell UE3) are not close to any HeNB, they share the rest of the frequency spectrum along with the femtocell UEs (i.e., femtocell UE1, femtocell UE2, and femtocell UE3). However, this scheme does not consider inter-HeNB interference and may be inefficient if the number of macrocell UEs near the HeNB increases.

2.3.2 Clustering of Small Cells

In [3], a framework is presented to reduce DL interference (both cross-tier and co-tier) and enhance the spectral efficiency for an OFDMA-based closed access femtocell network. In this framework, a Femtocell System Controller (FSC) per macrocell obtains all the necessary knowledge of HeNB system configuration (i.e., position information of HeNBs and macrocell UEs) and performs the necessary computations. To mitigate interference, the scheme encompasses a combination of dynamic frequency band allocation among HeNBs and MeNB, and clustering of HeNBs based on their geographical locations. In this scheme, a portion of the entire frequency band is dedicated to the MeNB users and the rest is reused by the MeNB and HeNBs. The advantage of allocating a portion of the frequency band strictly for MeNB users is that it can solve the MeNB UE DL deadzone problem and guarantee users’ QoS requirement. However, the portion of the frequency band to be shared, is determined by the total number of HeNB clusters obtained through a clustering algorithm.

The clustering algorithm allocates HeNBs into different frequency reuse clusters and UEs of different HeNBs in the same cluster use the same sub-channels allocated from the shared frequency band. Based on the geographical locations of the HeNBs, the threshold distance for clustering is calculated. If the Euclidean distance between any two HeNBs is less than the threshold distance, then they are assigned to different clusters to avoid co-tier interference and CTI. Simulation results show that high spectrum efficiency is achieved as the probability of cross-tier spectrum reuse becomes high. This signifies that the problem regarding a macrocell UE DL deadzone around HeNBs is effectively solved. For the proposed scheme, simulation results also show a significant improvement of the femtocell user capacity.

In [4], an energy-efficient interference mitigation scheme is presented for closed access HeNBs grouped in a neighborhood area based on their geographical locations. In this scheme, inter-femtocell or co-tier interference among neighboring HeNBs is minimized by reducing the unnecessary Available Intervals (AI) in Low Duty Operation (LDO) mode for HeNBs, which is determined based on the pattern of Low Duty Cycle (LDC). According to the IEEE 802.16m standard, an HeNB in the operation state may enter the LDO mode if no UE exists in its coverage zone, or if all the UEs in the coverage are in sleep/idle mode. In the LDO mode, an HeNB switches alternately between available interval (AI) and unavailable interval (UAI) modes. During UAI, an HeNB becomes inactive on the air interface. During AI, the HeNB may become active on the air interface by transmitting preambles to the new incoming UE for synchronization purposes. However, the HeNB in the LDO mode still has AIs even though there is no UE that will access the HeNB in near future. These unnecessary AIs cause co-tier interference for CGS HeNBs. In the proposed scheme, the unnecessary AIs are reduced which results in reduction of co-tier interference among neighboring HeNBs. The main idea behind reducing such interference is to cluster/group the neighboring HeNBs based on their geographical locations. In each cluster, one HeNB is designated as the leader and the adjacent HeNBs are referred to as members.

According to the IEEE 802.16m standard, a newly installed HeNB scans the surrounding area to search for neighboring HeNBs in its initialization state. Since it is assumed that the global knowledge about the topology of the network is available, the scanning report may include the group configuration in the network (i.e., the leader and the members of the group based on the HeNB ID). If a newly installed HeNB receives the preamble signal from the leader above a defined threshold then it becomes a member of the group. Otherwise, the newly installed HeNB will form a new group and assign itself as the leader of the group. The leader requires AIs in its LDC pattern so that the arrival of a UE at the group can discover the existence of the group by detecting the leader, even though the members in the group stay in UAI. As soon as the leader senses the arrival of the UE, it sends a message to the target HeNB to activate its AI in the LDC pattern so that the UE can detect the target HeNB and connect to it. In this pattern, unnecessary AI in the LDO mode of HeNB is reduced resulting in power conservation of HeNB and at the same time the co-tier interference is minimized. Through analysis and simulation, it is shown that for the proposed scheme, the gain in terms of co-tier interference reduction time and energy saving is up to 90% in comparison with conventional LDO scheme in the IEEE 802.16m standard.

In [5], another clustering-based scheme is proposed to mitigate the intra-tier interference among CA-based femtocells. Note that in the LTE-Advanced system, two or more CCs belonging to contiguous or non-contiguous frequency bands can be aggregated to support a wider transmission bandwidth. Femtocells that are likely to have an interfering relationship are grouped into the same cluster and femtocells in the same cluster are allowed to exchange local information with each other. In the proposed scheme, for selection of CC, the femtocells consider interference experienced in that CC, and they cooperatively adjust their transmission power in order to optimize the total throughput on each CC.

2.3.3 Beam Subset Selection Strategy

The authors in [6] propose an orthogonal random beamforming-based CTI reduction scheme in closed access two-tier femtocell networks. The macrocell beam subset selection strategy is based on the number of macrocell UEs and the intensity of HeNBs in the network. The MeNB selects the beam subset and the users for each channel based on the signal-to-interference-plus-noise (SINR) information for all the channels which is fed back by the macrocell UEs. The main objective is to enhance network throughput by optimizing the trade-off between multiplexing gain and multiuser interference (cross-tier) based on adaptive selection of optimal number of beams using max-throughput scheduler at the MeNB. The adaptive selection of the number of beams decreases CTI, and provides spatial opportunity to HeNBs to access the spectrum in an opportunistic manner. In addition, distributed power control mechanism for HeNBs integrated with the proposed scheme reduces CTI significantly.

2.3.4 Collaborative Frequency Scheduling

Co-channel UL and DL CTI can be mitigated if an HeNB can avoid using the macrocell RBs that belong to its nearby macrocell UEs through efficient spectrum sensing. However, the spectrum sensing results for HeNB may be impaired due to misdetection, false alarm, and improper timing synchronization. To deal with this problem, a framework for OFDMA-based HeNBs is provided in [7] where the scheduling information for macrocell UEs’ (both UL and DL) is obtained from the MeNB through a backhaul or air interface. This information is used to improve the spectrum sensing results for HeNB and to utilize the RBs associated with a far-away macrocell UE in the UL and DL transmissions. The key features of the proposed framework are as follows.

• HeNB receives the macrocell UEs scheduling information for UL and DL transmissions from the MeNB.
• HeNB performs spectrum sensing for finding the occupied parts of the spectrum. The occupied parts of the UL spectrum can be determined through energy detection.
• HeNB compares the spectrum sensing results with the obtained scheduling information to decide about the spectrum opportunities.

Since the HeNB accesses the spectrum in an opportunistic manner, the authors analyze the impact of Inter-carrier Interference (ICI) from macrocell UEs to femtocells which is severe in the UL transmission. The ICI is basically due to the asynchronous arrival of macrocell UE signals at the femtocell. Through simulations (using Okumura–Hata model of radio propagation), it is shown that the variation of ICI power depends on center frequency, height of the femtocell, and length of the Cyclic Prefix (CP). A lower center frequency and a higher femto BS height increase the received ICI power at the HeNB. In addition, if the macrocell UEs’ signal arrival time at HeNB exceeds the CP duration then the orthogonality between the sub-carriers is disrupted leading to ICI. Also, different sub-carrier assignment schemes result in different ICI.

2.3.5 Power Control Approach

Power control methods for CTI mitigation generally focus on reducing transmission power of HeNBs. These methods are advantageous in that the MeNB and HeNBs can use the entire bandwidth with interference coordination. Dynamic or adjustable power setting, which is preferred over fixed HeNB power setting, can be performed either in proactive or reactive manner and each of which again can be executed either in open-loop power setting (OLPS) or closed-loop power setting (CLPS) mode. In the OLPS mode, an HeNB adjusts its transmission power based on its measurement results or predetermined system parameters (i.e., in a proactive manner). In the CLPS mode, the HeNB adjusts its transmission power based on the coordination with MeNB (i.e., in a reactive manner). Also, a hybrid mode can be used where the HeNB switches between the two modes according to the operation scenarios [9].

Another related concept is power control for HeNBs on a cluster basis in which the initial power setting for the HeNBs is done opportunistically based on the number of active femtocells in a cluster (Figure 2.3) [8]. For this, centralized sensing can be used by which an MeNB can estimate the number of active femtocells per cluster and broadcast the interference allowance information to femtocells for their initial power setting. Alternatively, distributed sensing can be used where each cell senses if the others are active in the same cluster and adjusts its initial power setting accordingly.

In [10], a joint frequency bandwidth dynamic division, clustering and power control algorithm (JFCPA) for DL transmission is proposed to control both intra-tier and inter-tier interferences and to improve the system efficiency. The overall system bandwidth is divided into three bands. The macro-cellular coverage is divided into two areas taking the intensity of the interference from the macro base station to the femtocells into account. Both of these are dynamically computed using the JFCPA. A cluster is defined as the unit for frequency reuse among femtocells. The problem of clustering is mapped to the MAX k-CUT problem to eliminate the inter-femtocell collision interference and a graph-based heuristic algorithm is developed to solve it. Frequency bandwidth sharing or splitting between the femtocell tier and the macrocell tier is determined by a step-migration-algorithm (SMA)-based power control method which allocates the minimal transmission power to every FAP such that the FUEs’ SINR requirement is satisfied in any position of the femtocell.

For a single-carrier two-tier cellular system, in [11], the authors presented an estimation-based distributed UL power allocation approach for the FUEs to maximize their capacity in a fair manner. The UL power allocation problem for the FUEs in a macrocell–femtocell network can be modeled as a non-linear program (NLP) with an objective to maximizing the weighted sum-capacity of FUEs in different femtocells subject to the individual maximum power constraint at each FUE. However, this NLP can only be solved in a centralized manner. Therefore, the authors propose a distributed method in which, in each time slot, in each femtocell, the active FUE selects its power based on interference prediction and the SINR evolution. There is no exchange of any information among the FAPs, and among FAPs and macro base station. At every time slot, each FAP reports to its active FUE some interference related measurements to help the FUE select the transmission power.

Game theoretic models can be used to design and analyze distributed power control methods in a heterogeneous cellular wireless network with macrocells and femtocells. Two broad categories of game theoretic models are noncooperative and cooperative game models. In [12], a distributed power control problem is formulated for DL transmission in OFDMA-based femtocells overlaid upon a macrocell network. The problem is modeled as a noncooperative game, namely, a Stackelberg game, where the throughput of each station in the network is maximized under power constraints. In this game, the macrocell UEs are referred as the leaders and the femtocell UEs are considered to be the followers. The game is divided into two sub-games: a sub-game comprised of the set of leaders, referred to as the upper sub-game, and another sub-game comprised of the set of followers, referred to as the lower sub-game. The players in each sub-game compete with each other in a noncooperative manner to reach a sub-game Nash equilibrium, which is the solution of the power control game.

In [13], the EE problem for green communications is solved by using a power allocation approach to maximize BS EE when macrocells and femtocells share the same frequency spectrum. The telecom energy efficiency ratio (TEER) is defined as the ratio between effective system throughput and energy consumption of both macro BS and FAPs. An optimization problem is formulated to obtain power allocation for DL transmission in order to maximize local TEER while considering the reference FUE and the closely located interfered MUE (i.e., the CTI). Due to non-convex nature of the problem, a local optimum value is obtained by using the interior-point method for nonconvex NLPs.

2.3.6 Cognitive Radio Approach

Cognitive radio approach based on distributed spectrum sensing can be used for interference mitigation in femtocell networks. In [14], an efficient DL co-tier interference management scheme for an OFDMA-based LTE system is proposed where the path-loss information is shared among HeNB neighbors. In addition, adjacent HeNBs share the information related to the usage of LTE CCs, which are formed based on CA, in a distributed manner. The exchange of information between HeNBs may be done via femtocell gateway (HeNB GW) or over-the-air (OTA). The HeNB GW is an intermediate node between HeNBs and mobile core network that manages the inter-HeNB coordination messages via the S1 connection. On the other hand, the OTA method includes a direct link between HeNB and MeNB.

In the proposed scheme, when an HeNB is turned on, it identifies the adjacent neighbors and obtains the knowledge of the CCs used by the neighbors. The main idea of the scheme is that, each HeNB estimates the co-tier interference based on the path-loss information, capitalizes the knowledge of the usage of CCs by the neighbors, and accesses the spectrum intelligently to minimize interference. The selection of CC is done in such a way that, each HeNB selects the CC which is not used by the neighbor, or the CC that is occupied by the furthest neighbor, or the CC that is occupied by the least number of neighbors (in a chronological order as mentioned). Simulation results show a significant reduction in co-tier interference and signaling overhead within the network when compared with another cognitive based HeNB co-tier interference management technique.

Figure 2.4 illustrates a scenario of co-tier interference management (DL) of HeNBs through cognitive approach. In this scenario, the available CCs for HeNBs are CC1, CC2, CC3, and CC4. In Figure 2.4(a), since HeNB1 and HeNB3 are adjacent to each other, they select different CCs. On the other hand, since HeNB2 is a neighbor of neither HeNB1 nor HeNB3, it selects any one pair of available CCs (e.g., CC1 and CC2). Now, under such femtocell deployment, when HeNB4 is turned on, it discovers its adjacent neighbors, that is, HeNB1 and HeNB2. Through inter-HeNBs coordination mechanism, HeNB4 obtains the information related to the usage of CCs of its adjacent neighbors. Thus, in order to avoid co-tier interference, when HeNB4 selects CCs for the DL transmission, it selects the CCs (i.e., CC3 and CC4) which are different from those used by HeNB1 and HeNB2. Furthermore, in Figure 2.4(b), when HeNB5 is turned on, it identifies the adjacent neighbors (i.e., HeNB1, HeNB3, and HeNB4) and obtains the knowledge of the CCs used by the neighbors. Under these circumstances, HeNB5 selects the CCs that are occupied by the furthest neighbor, that is, HeNB1. To this end, HeNB5 selects CC1 and CC2 for DL transmission to reduce co-tier interference.

2.3.7 Fractional Frequency Reuse (FFR) and Resource Partitioning

The basic mechanism of this method divides the entire frequency spectrum into several sub-bands. Afterward, each sub-band is differently assigned to each macrocell or sub-area of the macrocell. Since the resources for the MeNB and HeNBs are not overlapped, interference between MeNB and HeNB can be mitigated.

In [15], the authors propose a frequency sharing mechanism that uses frequency reuse coupled with pilot sensing to reduce cross-tier/co-channel interference between macrocell and femtocells. In this scheme, FFR of three or above is applied to the macrocell. When an HeNB is turned on, it senses the pilot signals from the MeNB and discards the sub-band with the largest received signal power, and thus uses the rest of the frequency sub-bands resulting in an increased SINR for macrocell UEs. The overall network throughput is enhanced by adopting high order modulation schemes.

In [16], another interference management scheme for LTE femtocells is presented based on FFR. The scheme avoids DL CTI by assigning sub-bands from the entire allocated frequency band to the HeNBs that are not being used in the macrocell sub-area. In the proposed scheme, the macrocell is divided into a center zone (corresponding to 63% of the total macrocell coverage area) and an edge region including three sectors per each region. The reuse factor of one is applied in the center zone, while the edge region adopts the reuse factor of three. The entire frequency band is divided into two parts and one of them is assigned to the center zone. The rest of the band is equally divided into parts and assigned to the three edge regions.

Figure 2.5 illustrates the allocation of frequency sub-bands within the macrocell sub-areas. The sub-band A is used in the center zone (Cl, C2, and C3), and sub-bands B, C, and D are used in regions X1, X2, and X3, respectively. Now, when an HeNB is turned on, it senses the neighboring MeNB signals, compares the Received Signal Strength Indication (RSSI) values for the sub-bands, and chooses the sub-bands which are not used in the macrocell sub-area. In addition, if the HeNB is located in the center zone, then it excludes the sub-band that is used in the center zone as well as the one that is used by the macrocell in the edge region of the current sector. For example, if an HeNB is located in edge region X1, then it would exclude sub-band B which is used by macrocell UEs, and select sub-band A, C, or D. However, if an HeNB is located in center zone C1, then it avoids sub-band A and at the same time sub-band B since the received signal strength indicator (RSSI) for this sub-band is comparatively higher for that HeNB. In this way, this scheme mitigates co-tier interference and CTI. Simulation results show that, the scheme offers throughput gains of 27% and 47% on average, when compared with the FFR-3 scheme (with no center zone) and a scheme with no FFR, respectively.

To improve spectrum utilization when compared to that for orthogonal spectrum assignment for macrocell and femtocells, the authors in [17] use partially shared spectrum which is a solution between orthogonal and shared assignments. In this scheme, the macrocell and femtocell tiers access their own spectrum bands which overlap partially with each other.

The schemes described above use a fixed partitioning, which would cause a loss in throughput performance due to inefficient use of the bandwidth resources. A dynamic partitioning scheme (in both time and frequency domain) can be used for bandwidth sharing which minimizes CTI. In [18], an adaptive FFR scheme is presented to minimize DL interference caused by the HeNBs in the vicinities of a macrocell. The proposed scheme adopts FFR radio resource hopping or orthogonal FFR radio resource allocation based on the density (e.g., high or low) and location information (e.g., inner region or outer region) of the HeNBs. The location information of the HeNBs may be obtained and maintained within the network through using registered physical address associated with the broadband IP address that an HeNB uses. The proposed scheme only deals with the CTI posed by the HeNBs located (inner region) near the MeNB. If the HeNB is situated in a highly dense inner region, then orthogonal sub-channels are adopted by the HeNBs. Otherwise, the HeNB selects a sub-channel arbitrarily, utilizes it for a certain period of time, and then hops to other sub-channels. The proposed scheme reduces DL CTI.

The effect of spectrum allocation in two-tier networks is investigated in [19], where the macrocells employ closed access policy and the femtocells can operate in either open or closed access mode. By introducing a tractable model, the success probability for each tier under different spectrum allocation and femtocell access policies is derived. Both joint sub-channel allocation, in which the entire spectrum is shared by both tiers, as well as disjoint sub-channel allocation, whereby disjoint sets of sub-channels are assigned to both tiers, are considered. When femtocells are configured as closed access, each FAP is only accessible by its femtocell users, but when the femtocells are configured as open access, a FAP is accessed by both its users and all cochannel macro users. The throughput maximization problem is formulated subject to QoS constraints in terms of success probabilities and per-tier minimum rates.

Note that resource partitioning method can be used along with power control (thus resulting in a hybrid approach) to achieve co-tier interference and CTI. In [20], a joint sub-band, rate and power allocation scheme is proposed according to the interference statistics. Each femtocell determines its best sub-band, rate and power in a distributed manner in order to reach the maximum throughput on average.

2.3.8 Resource Scheduling Strategies

According to the different QoS requirements of the users, the resource scheduling schemes in OFDMA systems are generally classified into the following two categories: rate adaptive (RA) schemes and margin adaptive (MA) schemes [21, 22]. In an RA scheme, it is assumed that the overall transmission power is fixed and the objective is to allocate resources (sub-carrier and power) so that the total power does not exceed the overall transmission power limit. The fairness of resource allocation is achieved by using different techniques such as maximizing the minimum capacity among all users, using a proportional fairness constraint or using an objective function called utility function. In an MA scheme, each user has a fixed QoS demand (e.g., data rate and bit error rate [BER]) and the objective is to minimize the total transmission power while satisfying the QoS requirements of these users. MA-based schemes consider networks where all users have QoS demands.

Rate-adaptive scheduling schemes: In [23], a genetic algorithm is used to solve the resource allocation problem for macrocell–femtocell networks. The optimization of resource allocation is equivalent to the maximization of the total achievable throughput in the system, which is calculated using the Shannon’s formula subject to the constraints on total available bandwidth, power, and capacity in macrocell and femtocells. The algorithm uses two stages: BS selection and bandwidth and power allocation. Some users are allowed to be connected to femtocells together with the femtocell subscribers. A decision on BS selection is made for each user to determine whether the user gets access to the core network via a macrocell or a nearby femtocell. A subscriber is given priority to connect to its own femtocell while a non-subscribing user can join it if the user is located inside the femtocell’s coverage as long as the connection of the user does not affect the transmissions of the femtocell subscribers. After BS selection, a dual bandwidth and power assignment is implemented to distribute the available resources among users to maximize the overall system throughput. A portion of the available resource is allocated to each user depending on the demand and location of the user. The bandwidth assigned to each user in macro tier is less than or equal to demand but lower than the maximum allowed capacity in the zone the user is located at. In femto tier, bandwidth allocated to each user is kept less than or equal to the demand. The fitness function employed in the genetic algorithm is the objective function of the optimization problem which is performed after the random generation of the first population.

In [24], the resource allocation problem is formulated as a linear program (LP) which aims to maximize the system sum-rate under the proportional user rate constraint. The proportional rate constraint for users is regarded as a fairness criterion to balance the trade-off between the system sum-rate and user fairness.

A graph-based channel allocation scheme is proposed in [25] to maximize the system throughput while ensuring proportional fairness in rate among femtocells. The received signal-to-interference-plus-noise-ratio (SINR) is explicitly used by this scheme to generate the interference graph for the OFDMA femtocell network so as to guarantee the acceptable ICI for all the links. First, all the femtocells are partitioned into different groups by applying a greedy graph coloring algorithm to maximize the sum throughput of each group. The femtocells in the same group are given permission to share the assigned sub-channels while those in different groups are assigned orthogonal sub-channels. Then, an optimization problem is formulated to estimate the number of sub-channels assigned to each group. An approximation method is suggested to solve the optimization problem.

Margin-adaptive scheduling schemes: A distributed cell selection and resource allocation (CS–RA) mechanism is suggested in [26], where the CS–RA processes are performed by the UEs independently. The CS–RA problem is formulated as a two-tier game, namely, an inter-cell game and an intra-cell game. In the first tier, each UE selects the best cell according to its cell selection strategy which is stochastic, that is, there is a set of probabilities each representing the probability of selecting a particular cell. From the expected payoff, the optimal strategies of the UEs are derived, which depend not only on the channel qualities, but also on the load distribution of all cells and the strategies of other MSs. In the second tier, that is, in the intra-cell game, the UEs within the same cell choose the proper radio resource, typically sub-channels and power, to achieve their maximum payoff. Distributed algorithms, namely, the CS and RA algorithms are proposed to enable the independent UEs to converge to the Nash equilibria.

In [27], the available spectrum resource is dynamically used by each femtocell according to the macrocell resource usage. The proposed scheme gives priority to macro users and the resource allocation to the macro users is performed based on traditional optimization formulations. The concept of neighboring area around a femtocell is used which the macro users are vulnerable to CTI due to transmissions from the femto access points. A femtocell only uses the sub-carriers which are not used by the macro users inside the neighboring area. The resource allocation among the femtocells is modeled as a noncooperative game and a utility function is adopted that considers the benefits from sub-carrier and power allocation to the femto users.

In [28], a channel allocation scheme for two-tier OFDMA femtocell networks is recommended, where the soft frequency reuse (SFR) strategy is adopted by macro BSs. That is, a macrocell is divided into two regions: inner region and outer (or edge) region. For the MUEs located in the outer regions in macrocells, the frequency spectrum is equally partitioned across cells based on a reuse factor N. The MUEs in the inner regions of the macrocells use the sub-bands allocated to the outer regions of the neighboring cells. Considering that the macro user equipments (MUEs) have priority over the femtocell users (FUEs), the sub-bands are first assigned to MUEs utilizing SFR such that the ICI is mitigated. Subsequently, sub-bands are allocated to the femtocells such that CTI is mitigated. The scheme is optimized in terms of EE of the entire network (i.e., throughput per unit energy consumption), while guaranteeing that both the MUEs and FUEs attain at least a given minimum data rate.

A more detailed exposition to the resource scheduling problem in OFDMA networks will be provided in Chapter 3.

2.4 QUALITATIVE COMPARISON AMONG INTERFERENCE MANAGEMENT APPROACHES [30]

The “efficiency” of a scheme depends on whether it (i) mitigates/significantly reduces both co-tier interference and CTI; (ii) is applicable for both UL and DL transmissions; (iii) considers coordination among HeNBs and MeNB, or capitalizes on minimal amount of information, that is, path-loss, geographical location, or usage of the spectrum or sub-band among nearby HeNBs and/or among HeNBs and MeNB; (iv) handles ICI (e.g., by using frequency scheduling or any other method); (v) adopts an adaptive power control mechanism; (vi) corresponds to opportunistic access of the spectrum by the HeNBs based on RSSI value from MeNB signals; (vii) reduces the unnecessary AIs of LDO mode for HeNBs; (viii) is scalable and robust, that is, implementable for mass deployment of HeNBs; and (ix) is applicable for all three types of access modes (i.e., closed, open, and hybrid).

For example, the efficiency of cognitive approach can be considered to be moderate since it is capable of handling both CTI and co-tier interference with minimal amount of information (i.e., information about usage of sub-bands) exchange among neighboring HeNBs, applicable for all types of access modes of HeNBs, and more importantly, it accesses the spectrum in an opportunistic manner causing minimal harm to the nearby macrocell UEs. The collaborative frequency scheduling scheme can be considered to be highly efficient since it significantly reduces CTI and co-tier interference for mass deployment of HeNBs in both UL and DL transmission, handles ICI problem, and allows the HeNBs to opportunistically access the spectrum based only on the scheduling information of macrocell UEs that is exchanged between HeNBs and MeNB.

The “complexity” of each scheme increases with (i) the amount of information exchanged among neighboring HeNBs, (ii) the amount of information exchanged between HeNBs and MeNB, (iii) formation of clusters among HeNBs, (iv) algorithm executed in the HeNBs and/or in the MeNB to allow the HeNBs to access the spectrum opportunistically, etc. The more information exchanged among HeNBs or between HeNBs and MeNB, the more signaling overhead is introduced, and more processing is done in both HeNBs and MeNB, increasing the complexity of the scheme. For example, the complexity of the beam subset selection strategy scheme can be considered to be high since it requires the channel state information from all macrocell UEs to determine the optimal number of beams every time along with extensive coordination between HeNBs and MeNB regarding the spectrum access (thus increasing the signaling overhead). Also, the HeNBs have to run iterative power control algorithm to minimize interference.

Selection of an interference management scheme depends on the desired trade-off between complexity and efficiency. FFR can be considered to be a viable interference management scheme for two-tier femtocell networks due to the following reasons. First, the FFR requires minimal/no coordination among HeNBs and MeNB (and hence reduces the signaling overhead, and thus the complexity of the system), opportunistically accesses the spectrum based on only RSSI value from MeNB signals. Second, the FFR effectively solves the problem of CTI and co-tier interference in UL and DL transmissions for different access modes of HeNBs. Consequently, the FFR can increase the throughput of the network by a large margin, and can be used when the average number of HeNBs per macrocell is very high (about 180–200) while maintaining the QoS requirements of macrocell UEs. The FFR has been recommended as an effective interference management scheme for OFDMA-based two-tier networks [29].

2.5 FUTURE RESEARCH DIRECTIONS [30]

To enable mass deployment of femtocells, it is essential to develop distributed interference management schemes which primarily satisfy the QoS requirements of macrocell and femtocell UEs and at the same time enhance the capacity and coverage of the network. Such schemes should incur low overhead for coordination among macrocell BSs (i.e., MeNBs), and should also be able to integrate mobility management with different access modes and to address synchronization issues while keeping the complexity as minimal as possible. The interference management solution would strongly depend on the employed radio access technology (e.g., CDMA or OFDMA) and access mode (i.e., closed, open, or hybrid). In particular, adaptive admission control, power control, and advanced communication strategies such as interference cancellation and beamforming for multiple-antenna transceivers are important techniques to mitigate co-tier interference and CTI. For example, by using beamforming techniques femtocells can form antenna beams toward their UEs while nulling interference caused to macrocell UEs. In addition, macrocells would have higher priority in accessing the spectrum; therefore, suitable admission control mechanisms should be activated when femtocells create intolerable interference for macrocell UEs.

For OFDMA-based femtocell networks, if different sets of sub-channels are assigned to macrocells and femtocells, the CTI can be completely eliminated. However, to improve the spectrum utilization, a more efficient spectrum assignment method can be adopted.

Also, hybrid interference management schemes which combine power control with resource partitioning are promising. Power control schemes are advantageous in that MeNB and HeNB can use the entire bandwidth with interference coordination for both control and data channels. However, for this, the HeNB measurement scheme for power setting would need to be standardized. Also, such a scheme may not be fully effective when a macro UE is located very close to an HeNB. With resource partitioning schemes, interference between MeNB and HeNB can be eliminated. However, multiple frequency bands are required. The merits of both the approaches can be exploited in a hybrid scheme, the design of which is not trivial.

An important consideration in the design of resource management algorithms for multi-tier networks is minimizing the coordination among the small cells and that between the small cells and macrocells. Also, the limited capacity of the backhaul network needs to be considered.

Energy consumption in the cellular wireless networks has become a critical issue and EE has become a very important system design parameter. With the deployment of a large number of small cells, this EE issue will become even more significant. Therefore, the resource allocation algorithms for multi-tier networks should be designed to improve the EE of the overall network. For example, the resource allocation schemes can be designed for small cells such that the overall power consumption is minimized while satisfying the QoS requirements and maintaining fairness among users.

The provisioning of CA in the evolving LTE-Advanced systems adds another dimension to the resource allocation problem. With CA, two or more CCs can be aggregated to support transmission bandwidths up to 100 MHz. In a multi-tier network, the RBs corresponding to the different CCs can be shared among the macro and the small cell users considering the transmission characteristics of the different CCs as well as the CA capabilities of the users (e.g., due to power limitations at the UEs). Effective resource allocation schemes need to be designed for these scenarios.

REFERENCES

1. M. Salem, A. Adinoyi, H. Yanikomeroglu, and D. Falconer, “Opportunities and challenges in OFDMA-based cellular relay networks: A radio resource management perspective,” IEEE Transactions on Vehicular Technology, vol. 59, no. 5, pp. 2496–2510, 2010.

2. W. Yi, Z. Dongmei, J. Hai, and W. Ye, “A novel spectrum arrangement scheme for femtocell deployment in LTE macrocells,” in Proceedings of IEEE 20th Symposium on Personal, Indoor and Mobile Radio Communications, pp. 6–11, 13–16 September 2009.

3. H. Li, X. Xu, D. Hu, X. Qu, X. Tao, and P. Zhang, “Graph method based clustering strategy for femtocell interference management and spectrum efficiency improvement,” in Proceedings of IEEE 6th International Conference on Wireless Communications Networking and Mobile Computing (WiCOM), pp. 1–5, 23–25 September 2010.

4. H. Widiarti, S. Pyun, and D. Cho, “Interference mitigation based on femtocells grouping in low duty operation,” in Proceedings of IEEE 72nd Vehicular Technology Conference Fall (VTC’10-Fall), pp. 1–5, 6–9 September 2010.

5. B. Wang, Y. Zhang, W. Wang, M. Lei, and L. Jiang, “A cooperative downlink power setting scheme for CA-based femtocells,” in Proceedings of 2012 IEEE 75th Vehicular Technology Conference (VTC-Spring), 2012.

6. S. Park, W. Seo, Y. Kim, S. Lim, and D. Hong,“Beam subset selection strategy for interference reduction in two-tier femtocell networks,” IEEE Transactions on Wireless Communications, vol. 9, no. 11, pp. 3440–3449, November 2010.

7. M. E. Sahin, I. Guvenc, Moo-Ryong Jeong, H. Arslan, “Handling CCI and ICI in OFDMA femtocell networks through frequency scheduling,” IEEE Transactions on Consumer Electronics, vol. 55, no. 4, pp. 1936–1944, November 2009.

8. M. S. Jin, S. Chae, and D. I. Kim, “Per cluster based opportunistic power control for heterogeneous networks,” in Proceedings of IEEE VTC’11-Spring, Budapest, Hungary, May 2011.

9. 3GPP R1-105238, “Further Discussion on HeNB Downlink Power Setting in HetNet,” 3GPP RAN1 Meeting, Xian, China, October 2010.

10. H. Li, X. Xu, D. Hu, X. Tao, P. Zhang, S. Ci, and H. Tang, “Clustering strategy based on graph method and power control for frequency resource management in femtocell and macrocell overlaid system,” Journal of Communications and Networks, vol. 13, no. 6, December 2011, pp. 664–677.

11. N. Chakchouk and B. Hamdaoui, “Estimation-based non-cooperative power allocation in two-tier femtocell networks,” in Proceedings of 2011 IEEE Global Telecommunications Conference (GLOBECOM 2011), 5–9 December, 2011, pp. 1–6.

12. S. Guruacharya, D. Niyato, D. I. Kim, and E. Hossain, “Hierarchical competition for downlink power allocation in OFDMA femtocell networks,” IEEE Transactions on Wireless Communications, vol. 12, no. 4, pp. 1543–1553, April 2013.

13. B. Han, W. Wang, and M. Peng, “A power allocation scheme for achieving high energy efficiency in two-tier femtocell networks,” in Proceedings of 2011 IEEE 13th International Conference on Communication Technology (ICCT), 2011.

14. L. Zhang, L. Yang, and T. Yang, “Cognitive interference management for LTE-A femtocells with distributed carrier selection,” in Proceedings of IEEE 72nd Vehicular Technology Conference Fall (VTC 2010-Fall), pp. 1–5, 6–9 September 2010.

15. T. Kim and T. Lee, “Throughput enhancement of macro and femto networks by frequency reuse and pilot sensing,” in Proceedings of IEEE International Performance, Computing and Communications Conference (IPCCC), pp. 390–394, December 2008.

16. L. Poongup, L. Taeyoung, J. Jangkeun, and S. Jitae, “Interference management in LTE femtocell systems using fractional frequency reuse,” in Proceedings of 12th International Conference on Advanced Communication Technology, vol. 2, pp. 1047–1051, 7–10 February 2010.

17. L. Yang, L. Zhang, T. Yang, and W. Fang, “Location-based hybrid spectrum allocation and reuse for tiered LTE-A networks,” in Proceedings of IEEE 73rd Vehicular Technology Conference (VTC Spring), 2011.

18. R. Juang, P. Ting, H. Lin, and D. Lin, “Interference management of femtocell in macro-cellular networks,” in Proceedings of Wireless Telecommunications Symposium (WTS’10), pp. 1–4, 21–23 April 2010.

19. W. C. Cheung, T. Q. S. Quek, and M. Kountouris, “Throughput optimization, spectrum allocation, and access control in two-tier femtocell networks,” IEEE Journal on Selected Areas in Communications, vol. 30, pp. 561–574, 2012.

20. Z. Zhang, K. Wu, and A. Huang, “Optimal distributed subchannel, rate and power allocation algorithm in OFDM-based two-tier femtocell networks,” in Proceedings of IEEE 71st Vehicular Technology Conference (VTC 2010-Spring), 16–19 May 2010, pp. 1–5.

21. S. Bashar and Z. Ding, “Efficient algorithms for resource allocation in heterogeneous OFDMA networks,” in Proceedings of IEEE Global Telecommunications Conference (Globecom), November 30–December 4, 2008.

22. C. Liu, A. Schmeink, and R. Mathar, “Dual optimal resource allocation for heterogeneous transmission in OFDMA systems,” in Proceedings of IEEE Global Telecommunications Conference (Globecom), November 30–December 4, 2009.

23. H. Otrok, H. Barada, R. Estrada, A. Jarray, and Z. Dziong, “Resource allocation in macrocell-femtocell network using genetic algorithm,” in Proceedings of 2012 IEEE 8th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), 2012.

24. P. Xue, P. Gong, J. H. Park, D. Park, and D. K. Kim, “Radio resource management with proportional rate constraint in the heterogeneous networks,” IEEE Transactions on Wireless Communications, vol. 11, pp. 1066–1075, 2012.

25. K. Zheng, Y. Yang, C. Lin, X. Shen, and J. Wang, “Graph-based interference coordination scheme in orthogonal frequency-division multiplexing access femtocell networks,” IET Communications, vol. 5, pp. 2533–2541, 2011.

26. L. Gao, X. Wang, G. Sun, and Y. Xu, “A game approach for cell selection and resource allocation in heterogeneous wireless networks,” in Proceedings of 2011 8th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks (SECON), 2011, pp. 530–538.

27. S. Han, B.-H. Soong, and Q. D. La, “Interference mitigation in resource allocation for OFDMA-based macro/femtocell two-tier wireless networks,” in Proceedings of 2012 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB), 2012, pp. 1–6.

28. W. Li, W. Zheng, H. Zhang, T. Su, and X. Wen, “Energy-efficient resource allocation with interference mitigation for two-tier OFDMA femtocell networks,” in Proceedings of 2012 IEEE 23rd International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC), 2012.

29. Femto Forum, available at: http://www.femtoforum.org/femto. (http://www.smallcellforum.org)

30. 3GPP R1-106052, “Per Cluster Based Opportunistic Power Control,” 3GPP RAN1 Meeting, Jacksonville, USA, November 2010.

1. The MUEs are considered as outdoor users and the FUEs are considered as indoor users.