科研之友
微信
新浪微博
Facebook
分享链接
摘要
We study multicast capacity for a large-scale spatial inhomogeneous mobile network consisting of n ad hiac nodes. Under our mobility model, the stationary spatial distribution of a node is non-uniform; each node spends most of the time in a certain region, and rarely (or never) visits out of such region. To characterize the inhomogeneity of the mobility model, we define an activity exponent gamma and two clustering parameters (m(n),r(n)), where gamma is an element of [0,1] measures the strength of node mobility, m(n) denotes the number of clusters, r(n) denotes the radius of the cluster. We classify the mobility into two cases according to the strength of mobility of each node, called strong and weak mobility, respectively. Two corresponding scheduling schemes and routing policies combined with the Manhattan multicast tree method are proposed. Suppose there are n(s) = Theta(n) multicast sessions. Each source has n(d) destinations which are selected randomly and independently. We show that under strong mobility case, the per-node multicast capacity is (1/root n(d)theta(n)) with theta(n) = n1-o/2; under weak mobility case, when n(d) = O(m(n)/log m(n)), the multicast throughput is (1/root n(d)root m(n)/n(2) log m(n)); when n(d) = (m(n)/log m(n)) the multicast throughput is (1/n). Particularly, as a special case, i.e., by letting n(d) = 1, our results unify the previous unicast capacity bounds.
- 出版日期2013-1
- 单位同济大学
