Impacts of Topology and Traffic Pattern on Capacity of Hybrid Wireless Networks

In this paper, we investigate the throughput capacity in wireless hybrid networks with various network topologies and traffic patterns. Specifically, we consider n randomly distributed nodes, out of which there are n source nodes and n^d (0 < d < 1) randomly chosen destination nodes, together with n^b (0 < b < 1) base stations in a network area of [0, n^w] times [0, n^1-w] (0 < w les 1/2 ). We first study the throughput capacity when the base stations are regularly placed and their transmission power is large enough for them to directly transmit to any nodes associated with them. We show that a per-node throughput of max{min{n^b-1, n^d-1}, min {n^w-1/radiclog n, n^d-1}} bits/sec is achievable by all nodes. We then investigate the throughput capacity when the base stations are uniformly and randomly placed, and their transmission power is as small as that of the normal nodes. We present that each node can achieve a throughput of max{min{n^{b-1/log n}, n^d-1}, min {n^w-1/radiclog n, n^d-1}} bits/sec. In both settings, we observe that only when d > b and d > w, the maximum achievable throughput can be determined by both the number of base stations and the shape of network area. In all the other cases, the maximum achievable throughput is only constrained by the number of destination nodes. Moreover, the results in these two settings are the same except for the case d > b > w, in which the random placement of base stations will cause a degradation factor of log n on the maximum achievable throughput compared to the regular placement. Finally, we also show that our results actually hold for different power propagation models.