Prediction of Nearest Neighbor for Effective Data Transmission using Machine Learning

Abstract

In this paper we describe a new multicast routing protocol for mobile ad hoc wireless networks. The protocol establishes a source based mesh of nodes called with no static periphery or zone to distribute data for that source.  The notion of   zone or a _flooding group is different  the flooding group is created based on hop count distance metrics and distance constraints . Also the zone or flooding group is a source based mesh. The protocol aims to improve connectivity and data delivery amidst topology changes and node movement. It avoids the drawbacks of tree based protocols in ad hoc networks viz fragility against topology changes, non-optimal paths in the case of shared trees, tree partitions, frequent tree reconstruction etc. Also the protocol avoids excessive redundant data transmission due to multiple paths by using probabilistic data forwarding. Thus this protocol attempts to combine the robustness of the mesh structure by establishing Dynamic periphery  and improved efficiency by using a probabilistic data forwarding.

This is an on-demand protocol i.e.; control messages are distributed only when the source has data to send, thereby reducing channel overhead. The protocol uses a soft-state approach to maintain multicast group membership. The members do not send explicit messages to leave the group. The protocol is independent of the underlying unicast routing protocol.