Virtual Full Duplex Wireless Broadcasting via Compressed Sensing (1101.0294v3)

Abstract: A novel solution is proposed to undertake a frequent task in wireless networks, which is to let all nodes broadcast information to and receive information from their respective one-hop neighboring nodes. The contribution is two-fold. First, as each neighbor selects one message-bearing codeword from its unique codebook for transmission, it is shown that decoding their messages based on a superposition of those codewords through the multiaccess channel is fundamentally a problem of compressed sensing. In the case where each message consists of a small number of bits, an iterative algorithm based on belief propagation is developed for efficient decoding. Second, to satisfy the half-duplex constraint, each codeword consists of randomly distributed on-slots and off-slots. A node transmits during its on-slots, and listens to its neighbors only through its own off-slots. Over one frame interval, each node broadcasts a message to neighbors and simultaneously decodes neighbors' messages based on the superposed signals received through its own off-slots. Thus the solution fully exploits the multiaccess nature of the wireless medium and addresses the half-duplex constraint at the fundamental level. In a network consisting of Poisson distributed nodes, numerical results demonstrate that the proposed scheme often achieves several times the rate of slotted ALOHA and CSMA with the same packet error rate.