Neural Network´s k-means Distance-Based Nodes-Clustering for Enhanced RDMAR Protocol in a MANET

k-means distance-based nodes clustering technique proposed enhance the performance of RDMAR protocol in a Mobile Ad-hoc NETwork (MANET). To limit the flood search to just a circular local area around the source, the Relative Distance Micro-discovery Ad Hoc Routing (RDMAR) protocol uses the Relative Distance (RD). If the distance of flood discovery is further limited by clustering the nodes with similar characters in to one group, different from the dissimilar characters´ group, the performance of the RDMAR implementation can be elevated. The k-means algorithm, similar to the one in unsupervised learning in pattern classification, can be recursively applied to re-classify the clusters as the MANET environment, resource availability, and node demands change. This technique can be more effective in a MANET with comparatively moderate change of the dynamicity and slow change in nodes´ demands plus highly accumulated groups of nodes at given sub-areas.