An Adaptive Routing Framework for Load Balancing and Energy Efficiency in Wireless Sensor Networks

Abstract

Wireless networks that operate without fixed infrastructure face a persistent challenge in sustaining reliable Quality of Service, largely because bandwidth stays limited, links break without warning, and packets can be lost mid-transit. When a network falls back on standard recovery methods after such a disruption, the cost usually shows up as drained battery reserves and slower data delivery, and this cost compounds as the node count rises. This work responds to that challenge with LBCAR (Load-Balanced Congestion-Adaptive Routing), a protocol that pairs a Random Waypoint mobility model with real-time responsiveness to congestion forming at specific points in the network. Rather than triggering a complete route-rediscovery process whenever a path fails, LBCAR monitors three running indicators per node — how heavily loaded its traffic is, how much energy it has left, and how likely its current route is to hold — and uses that picture to reroute data through a locally mended path without pulling the original source back into the recovery process. Implementation and testing took place in NS-2, comparing LBCAR's behaviour against two established protocols, AOMDV and AOMDV-ER, across network sizes from 20 up to 100 nodes. The measurements collected show LBCAR maintaining higher throughput and a much tighter spread in energy usage than both comparison protocols, and end-to-end delay comes out lower too, provided node density stays in the low-to-moderate band.