Relay Selection Algorithms
Why consider relaying?
Future wireless networks will support relay-based communication, where well-placed relay nodes receive a message from a source node, process it, and forward it to its intended destination node. This will result in performance gains for end-users in systems that support the IEEE 802.11s and 802.16j standards. Relaying will be especially beneficial when there is no line-of-sight path between the source and the destination.
In dense wireless networks, there are typically several fixed relay nodes in the region between the source and the destination. Determining which of these potential relays should be selected is a difficult cross-layer problem. For example, a relay node may have a strong channel to the destination, but it may also be heavily loaded with traffic from other sources. Also, relays are usually battery-powered, and a potential relay may need to enter a sleep cycle to conserve energy.
The degree to which the source is involved in relay selection also cannot be discounted. Centralized selection strategies typically require a significant amount of overhead to be exchanged between the source and the relays. Thus, selecting relay nodes entails the resolution of various performance tradeoffs in order to maximize the overall quality of service.
Main results and future areas of research
We have proposed a decentralized relay selection strategy for two-hop networks. Our proposed approach relies on random access-based feedback to the source from relays that have decoded the message. If the destination detects uncorrectable packet errors, all decoding relays contend over a set of fixed minislots to send “Hello” messages to the source. At the end of the contention period, the source randomly chooses one of the relays that successfully sends a “Hello” message to it. We have also used channel feedback to improve the performance of our strategy by biasing relay selection towards the decoding relays with good channel gains to the destination.
We have also proposed centralized relay selection strategies for two-hop networks where transmission is based on a layered coding framework. Our objective is to maximize throughput by selecting a subset of the potential relays subject to a power constraint on the selected nodes. This is a difficult problem, so we approximate it by selecting the relays that are close to a particular location. This location-based approximation gives rise to two relay selection algorithms that yield throughputs close to those yielded by the optimal selection strategy.
We are currently investigating the connection between relay selection and user scheduling for the downlink of a cellular system. The presence of fixed relays in a cellular system requires that user scheduling decisions consider the ability of a relay to successfully deliver a packet to its intended mobile user, which would reduce the likelihood of buffer overflow at the base station.
Our initial results are summarized in:
 C.K. Lo, R.W. Heath, Jr., and S. Vishwanath, “The Impact of Channel Feedback on Opportunistic Relay Selection for Hybrid-ARQ in Wireless Networks,” submitted to the IEEE Trans. Veh. Technol., June 2007.