RELAY SELECTION AND POWER ALLOCATION IN WIRELESS COOPERATIVE NETWORKS

Abstract

Cooperative communication aims to achieve spatial diversity gain through cooperation of user terminals in transmission without requiring multiple antennas on the same node. It employs one or more terminals as relays in the neighborhood of the transmitter and the receiver, which collaborate in the transmission and serve as a virtual Multiple Input Multiple Output (MIMO) antenna array. Allowing cooperation in wireless communication introduces new problems related to resource allocation (like power, bandwidth) and relay selection. Relay selection is vital for reaping the performance benefits of cooperative communication. Relay selection is capable of extracting diversity with higher bandwidth efficiency as compared to distributed space-time coding. Relay selection mechanism may be centralized or distributed. A centralized mechanism employs a controlling unit which collects andutilizes the required channel information to select relays to each source-destination pair. In contrast to centralized relay selection, in a distributed mechanism, each node individually determines whether to cooperate or not based on the information exchange that occurs between nodes. In many relay selection schemes, the current observed channel condition is used to make the relay-selection decision for the subsequent frame. Since, the data rate is high (hundreds of kbits/s), the duration of a data block is smaller than the coherence time. When data transmission takes place, consecutive blocks of data experience approximately the same channel condition. To take the advantage of channel memory, the channel is approximated by means of a Markov model. We proposed a relay selection method using finite state Markov model for Rayleigh fading channel. It is aimed at achieving diversity gain and reducing signaling overheads. Further, Optimal Power Allocation (OPA) is applied along with the proposed relay selection method. It provides improvement in symbol error rate performance as expect