Energy Efficient Opportunistic Network Coding for Unicast Flows in Wireless Networks

Due to the broadcast nature of wireless channels, a packet sent by one node may be “overheard” by several nearby nodes. Rather than taking these overheard packets as interference, multiple-reception of packet can be exploited to improve network performance. In this paper, we consider energy efficient network coding design in wireless networks with multiple unicast sessions. Our approach decomposes multiple unicast sessions into a superposition of multicast and unicast sessions, with coding occurring only within each session. We give an optimizationbased approach that is more general than the existing poisonremedy optimization formulation. For the case of wireless, we consider XOR coding and give an achievable rate region for a primary interference model. To simplify network operation, we give an oblivious backpressure algorithm which may not utilize overhearing of transmissions optimally, and a practical protocol called COPR based on the oblivious backpressure algorithm. Simulation experiments show that COPR largely reduces network power consumption over existing algorithms.