Dynamic Resource Management for Coverage and Capacity Enhancement in Cellular Systems with Infrastructure Based Multihop Relaying

Dynamic resource management strategies addressing active set management of BS and relays, dynamic resource partitioning and dynamic routing are examined for use in cellular infrastructure based multihop relaying systems. Simulation results demonstrate improved coverage with full queue user data rate distribution, along with a small increase in throughput. The average web page download rate is also shown to improve significantly, reflecting a reduction in end-toend packet delay. Index Terms — Cellular Systems, Multihop, Resource Management Introduction Multihop relaying within the cellular infrastructure has been receiving considerable attention in WWRF [1] [2] as well as in recent literature [3] [5]. Multihop relays may be considered as a complement to the recent advances in air interface design and antenna processing technology which yield significantly higher spectral efficiencies than conventional 3G technologies. Multihop relays will enable the realization of these high spectral efficiencies over a wider area of the cell. Relays used in cellular networks may be infrastructure based, involving the use of fixed equipment to perform relaying or may leverage UEs within the network which are equipped with relaying functionality. Infrastructure based relaying has some advantages over User Equipment (UE) based relaying. Infrastructure based relaying requires the deployment of relay nodes, which implies an increase of infrastructure cost. However, it will minimize the additional enhancements in the UE, which are particularly significant in an FDD system, thus containing the UE cost as well as conserving its power. If relaying is done through the UE, every UE needs to have relaying capability. Also, the mobility of UEs would weaken the stability of the relay-UE link, and the coverage performance would depend on the density of UEs within the cell. In infrastructure based relaying, the nodes can be carefully located to address the need for hot spots serving high speed data users or the existence of deep shadowing spots within the cell. Furthermore, infrastructure based relays can be designed such that they can be added to a cell to enhance coverage without impacting legacy UEs, and only requiring some software enhancements in the access network. In [1], single hop conventional transmission and two-hop time-domain relaying are compared for different frequency reuse parameters, and it is shown that capacity and coverage gains are possible with simple two hop relaying, with increasing gains in larger reuse. A summary of multihop relaying research in WWRF projects is provided in [2]. Walke et al [3] propose a service architecture based on SIP for multihop relaying. Coverage enhancement in cellular networks with multihop relaying is addressed in [4]-[5]. Multihop relaying has also been examined for Hang Zhang, Derek Yu, Shalini Periyalwar Wireless Technology Labs, Nortel Networks, Ottawa Wireless World Research Forum (WWRF) Page 2 (8) range extension in a number of other systems, e.g., [6] [8]. Some important aspects of cellular infrastructure based multihop relaying have not been adequately covered in the literature. It is the objective of this paper to address dynamic resource management strategies in cellular infrastructure based multihop relaying for the following key areas, supported by simulation results for a system with nomadic users: 1. active set management 2. dynamic resource partitioning between BS-relay and relay-UE links 3. dynamic routing In a multihop cellular system where the cellular spectrum is being reused for the relay to UE hop, it is important to ensure that the resources are assigned efficiently. These resource management aspects are critical in ensuring that resources are allocated not only to get the highest net link rate for BS-UE communication, but also the lowest delay so as to enable support of services with different QoS requirements. The paper is structured as follows. Following a description of the system model used in the paper, dynamic resource management features are articulated. Active set management, dynamic resource partitioning and dynamic routing are addressed. Other resource management issues relating to relaying with mobility which are also discussed include efficient handling of MAC layer retransmissions and flow control during handoff. The performance improvements that the dynamic resource management schemes provide for nomadic users are demonstrated with end-to-end performance assessment for full-queue (FTP) and web browsing traffic. System Model An FDD based multi-beam[9] cellular system with universal (N=1) reuse is assumed, with relays deployed in all the beams in the system as shown in Figure 1. The multi-beam system may be considered equivalent to a multi-beam system, based on the antenna configuration. The focus of this work is on the downlink to the UE. The BSrelay and the relay-UE links employ the same frequency in a Time Division Multiplexed manner (Figure 1) where a fixed partitioning of resources between the BS-UE link, BSRelay link and Relay-UE link is illustrated. The physical layer is assumed to be OFDMA, such that multiple users could be supported in one transmission unit. Thus, the relay operates Rx from Relay IDLE Rx from BS UE Tx to UE Rx from BS IDLE Relay IDLE Tx to Relay Tx to UE Base 3 Relay to UE 2 BS to Relay 1 BS to UE Slot Figure 1 System model and resource assignment map with fixed partitioning for the cellular system with relays on the same frequency but in orthogonal TDM slots to communicate with the BS and UE respectively. Therefore, the UE is agnostic to whether the information was transmitted directly from the BS or through the relay. Proportionally fair scheduling [10] is assumed to be resident in the Base Station, where the relay transmissions slots are also defined. The relay incorporates a simple FIFO scheduler. Dynamic Resource Management Features With the objective of maximizing the performance of the cellular system with fixed relays, some dynamic resource management options are discussed below. Active Set Management Typically, the UE’s active set comprises Wireless World Research Forum (WWRF) Page 3 (8) the BSs with the strongest pilot measurement. Similarly, for the relays, the relay to BS association is determined by the best BS pilot seen by the relay (termed the relay’s parent BS). In order to find the best link to a UE however, the channel quality information on the BS-UE link and the relayUE link is required. Measuring the channel quality for the relay to UE link would require separate pilot signals for the relays. The active set could then comprise of both BS pilots and relay pilots. Then the parent beam for the UE can be chosen in one of two ways: based solely on the best pilot among BSs in the active set, or based on the best of all the BS and relay pilot measurements. In the latter case, it is possible that, for a given UE, if the relay pilot is the strongest of relay and BS pilots, the relay’s parent BS’s pilot may not necessarily be in the top of the UE’s active set. Then the question is whether it is more beneficial to associate the UE with the BS-relay pair for the best relay pilot or the BS-relay pair for the best BS pilot, for that UE. If the signal strength from the relay’s parent BS is adequate to support the control channel messaging to the UE, it makes perfect sense to assign the UE to the BSrelay pair for the best relay pilot. Otherwise, such a selection would require that the control signaling also be transmitted via the relay. Dynamic Resource Partitioning A simple approach to partitioning of resources in the multihop system is fixed partitioning, by allocating adjacent time slots to direct BS-UE transmission, BS-relay transmission and relay-UE transmissions, respectively, as shown previously in Figure 1. With dynamic resource partitioning, the resources are divided into two units and proportionally allocated to the BS-UE/relay transmissions and to relay-UE transmissions. The size of these units may be determined by the BS scheduler by taking into account the relay’s transmission needs based on the relay-UE link conditions as well as the BSUE/relay link conditions. Figure 2 illustrates an example of the dynamic resource partitioning scheme. It is possible to generalize this arrangement to have N BSRelay/UE slots and M Relay-UE slots. R2U Slot B2U Slot B2R Slot 1 slot 1 slot 1 slot