Statistically multiplexed adaptive operation of ad hoc networks with self-similar traffic

We have previously developed techniques for discovery, maintenance and adaptive updating of multiple routes in an ad hoc network with Poisson traffic and exponential service (packet transmission) times. In the present paper, we report simulation results using the same approach, but when the traffic is self-similar. The shapes of the performance curves (delay and buffer occupancy of queues) for Poisson and self-similar traffic are similar, especially in the middle range of loads; the curves are concave upwards, steeply rising, and look very similar. However the performance curves of queues with self-similar traffic saturate at smaller values, 60-80%, of the normalized load, in comparison with the results for queues with Poisson traffic. Results of simulation experiments with self-similar traffic demonstrate the following: even when the network topology is static, our algorithm responds to bursts in the traffic pattern and reduces buffer losses through the use of alternative, less congested routes. We also present simulation experiments to demonstrate the effect of simple changes in the network configuration.