Link state distribution rate scaling for maintaining topology in mobile wireless networks

In some current mobile ad hoc network designs, we have observed that the link state update protocol may take up all the network throughput and leave little to no capacity for data transmission. In this paper we explore the throughput scaling necessary to maintain an updated topology in a fixed rate mobile infrastructureless wireless network. We use a mobility model where each node travels along a random chord on the circle formed by the maximum transmission radius of the fixed rate system. Combined with a constant velocity and assuming independent movement, this allows us to determine a distribution for the time tos disconnection in the network, or link coherence time. For all-to-all topology updates, where after each link disconnect every node updates every other node of its link state, we show that the per node throughput for maintaining topologyscales as Ω(η2). We then develop numeric results for rate scaling using IP packets sized for Ethernet, that is 12 kbit (1.5 kbyte) packets. For this random chord mobility model, we show throughput scaling for networks of 10 to 1000 nodes. We plot results for 95%, 99%, and 99.9% probabilities of successful protocol distribution (based on the link coherence time distribution and the throughput) and for ratios of transmission range to velocity of 1, 10, and 100. We show that even for networks where the ratio of transmission range to velocity is 100 (that is, the transmission range is 100 times the velocity), the throughput scales on the order of Mb/s for networks with more than 100 nodes. For a 99.9% probability of successful protocol distribution, a network of 100 nodes with a ratio of transmission range to velocity of 100 requires a throughput per node of over 220 Mbps. For the same network with 1000 nodes, a throughput per node of over 23 Gbps is necessary.