Random Network Coding in Wireless Sensor Networks: Energy Efficiency via Cross-Layer Approach

The application of network coding (NC) in wireless sensor networks (WSNs), where each node broadcasts its information to all other nodes, is considered. Traditionally, this all-to-all broadcast is implemented by letting each node store and forward received packets. In the NC-based approach, each node overhears packets transmitted from neighbouring nodes, combines the received packets, and forwards the resulting packet to its neighbours. The hope is to need fewer transmissions, which helps to save energy. In this paper, we study the NC-based all-to-all broadcast assuming all inter-user channels are slow fading and reception energy is comparable to transmission energy. A cross-layer approach, that realizes the NC at the network layer and uses the MAC layer's forwarding and scheduling decisions to determine whether and how a packet has been network coded, is considered. Theoretical and simulation results show that the NC-based system consumes less energy than traditional store and forward (SaF) approach. However, consideration of non-zero receive energy reduces the energy gain of NC over SaF significantly; this is because more packets should be received to effectively utilize the network coding. Additionally, the performance of network coding over various field sizes is studied; results show that, for a fading channel, even smaller field sizes (e.g. F
₂
or F
₄
) perform well as compared to the traditional SaF approach.