On the Capacity of Random Wireless Networks Under Fixed Multipath Fading

We consider the problem of achievable per-node throughput in an extended distributed wireless network where the node locations are random and the channel attenuation between pairs of nodes exhibits independent random multipath fading. In [1] a clever protocol construction based on percolation theory was used to show that a per node throughput of a constant times 1/√n bps is achievable with probability approaching one as the expected number of nodes in network, n, becomes large (i.e. w.h.p.), for networks with random node locations and no multipath fading. We use a similar approach to extend the result to a more realistic channel gain model where the channel gains are random due to multipath fading effects. In particular, we show that when the channels are modeled as independent, frequency flat with Rayleigh distributed gains, a throughput of a constant times 1/√n log log n bps/node is achievable w.h.p..