Multi-constrained path (MCP) selection, in which the key objective is to search for feasible paths satisfying multiple routing constraints simultaneously, is known to be an NP-Complete problem. Multi-constrained path optimization (MCPO) is different from MCP mainly in that, the feasible paths selected should also be optimal with regard to an optimization metric, which makes path computation in MCPO even harder. We propose a fully distributed multi-constrained path optimization routing (MPOR)

more »

... tocol that solves the general ¤ -constrained path selection and routing optimization problems. MPOR computes paths using distance vectors exchanged only amongst neighboring nodes and does not require the maintenance of global network state about the topology or resources; supports hop-by-hop, connectionless routing of data packets, and implements constrained path optimization by distributively constructing an ¥ -optimal path set (i.e., the shortest, the second shortest and up to the ¥ th shortest path in terms of the optimization metric) for each destination at each node. Simulations show that MPOR has satisfactory routing success ratios for multiconstrained path selection, and performs consistently with varying number of constraints. For constrained path optimization, MPOR has high probabilities of finding feasible paths that are also optimal or near-optimal for the given optimization metric.