Analytical modeling of self-pruning and an improved probabilistic broadcast for wireless multihop networks

• Published in: Ad hoc networks 2016-12, Vol.52, p.106-116
• Main Authors: Rab, Raqeebir, Rahman, Ashikur, Zohra, Fatema Tuz
• Format: Article
• Language: English
• Subjects: Analytical modeling; Broadcast; Performance bounds; Self-pruning; Simson’s one-third rule
• ISSN: 1570-8705; 1570-8713
• DOI: 10.1016/j.adhoc.2016.08.007

The first wave of broadcasting algorithms exploit neighbor knowledge to reduce redundant retransmissions in mobile ad hoc wireless networks (MANETs). Self-pruning is one of such broadcast algorithm that received much attention from the research community. In self-pruning, only a subset of nodes forward the message based on certain forwarding rule. Moreover, it belongs to one of the reliable broadcasting algorithm category where a broadcast message is guaranteed (at least algorithmically) to reach all the nodes in the network. Despite its exclusive feature, so far the evaluation of self-pruning is mainly based on experiments and lacks detailed theoretical analysis. In this paper, we develop an analytical model for self-pruning to determine expected number of forwarding nodes required to complete a broadcast in the network. The derived expression is a function of various network parameters (such as, network density and distance between nodes) and radio transceiver parameters (such as transmission range). Moreover, the developed mathematical expression provides us a better understanding of the highly complex packet forwarding pattern of self-pruning and a valuable insight to design a new broadcasting heuristic. After careful analysis, we propose a new heuristic dubbed as dynamic probabilistic broadcast where the rebroadcast probability of each node is dynamically determined from a developed mathematical expression. Extensive simulation experiments have been conducted to validate the accuracy of the analytical model, as well as, to evaluate the efficiency of the proposed heuristic. Performance analysis shows that the proposed heuristic outperforms its static counterpart.