Search

A self-repelling random walk of a token on a graph is one in which at each step, the token moves to a neighbor that has been visited least often (with ties broken randomly). The properties of self-repelling random walks have been analyzed for two dimensional lattices and these walks have been shown to exhibit a remarkable uniformity with which they visit nodes in a graph. In this paper, we extend this analysis to self-repelling random walks on mobile networks in which the underlying graph itself is temporally evolving. Using network simulations in ns-3, we characterize the number of times each node is visited from the start until all nodes have been visited at least once. We evaluate under different mobility models and on networks ranging from 100 to 1000 nodes. Our results show that until about 85% coverage, duplicate visits are very rare highlighting the efficiency with which a majority of nodes in the network can be visited. Even at 100% coverage, the exploration overhead (the ratio of number of steps to number of unique visited nodes) remains low and under 2. Our analysis shows that self-repelling random walks are effective, structure-free tools for data aggregation in mobile ad-hoc networks.

This paper describes EZ-AG, a structure-free protocol for duplicate insensitive data aggregation in MANETs. The key idea in EZ-AG is to introduce a token that performs a self-repelling random walk in the network and aggregates information from nodes when they are visited for the first time. A self-repelling random walk of a token on a graph is one in which at each step, the token moves to a neighbor that has been visited least often. While self-repelling random walks visit all nodes in the network much faster than plain random walks, they tend to slow down when most of the nodes are already visited. In this paper, we show that a single step push phase at each node can significantly speed up the aggregation and eliminate this slow down. By doing so, EZ-AG achieves aggregation in only O(N) time and messages. In terms of overhead, EZ-AG outperforms existing structure-free data aggregation by a factor of at least log(N) and achieves the lower bound for aggregation message overhead. We demonstrate the scalability and robustness of EZ-AG using ns-3 simulations in networks ranging from 100 to 4000 nodes under different mobility models and node speeds. We also describe a hierarchical extension for EZ-AG that can produce multi-resolution aggregates at each node using only O(NlogN) messages, which is a poly-logarithmic factor improvement over existing techniques.

This paper describes Census, a protocol for data aggregation and statistical counting in MANETs. Census operates by circulating a set of tokens in the network using biased random walks such that each node is visited by at least one token. The protocol is structure-free so as to avoid high messaging overhead for maintaining structure in the presence of node mobility. It biases the random walks of tokens so as to achieve fast cover time; the bias involves short albeit multi-hop gradients that guide the tokens towards hitherto unvisited nodes. Census thus achieves a cover time of O(N/k) and message overhead of O(Nlog(N)/k) where N is the number of nodes and k the number of tokens in the network. Notably, it enjoys scalability and robustness, which we demonstrate via simulations in networks ranging from 100 to 4000 nodes under different network densities and mobility models., Comment: 25 pages, technical report, index terms:random walk, MANET, statistical aggregation, gossip, local gradients