Smart forwarding technique for routing with multiple QoS constraints

QoS-constrained routing is considered as one of the key components to support quality of service in next-generation data networks. However, the optimal routing problem subject to multiple constraints is NP-hard in general. In this paper we propose a technique called “smart forwarding” which can be used in both distributed hop-by-hop QoS routing and centralized source-based routing. It enables fast on-demand routing by utilizing a table pre-computed with link-state information or distributed Bellman-Ford algorithm. It can greatly reduce routing overhead in both flooding-based and crank-back routing protocols by only forwarding routing request to a neighbor that is known to be able or potentially be able to meet the QoS requirement. We also describe how we can adjust routing overhead by bounding the number of flooding or crank-back trials with this technique. More detailed analysis of this technique with delay-constrained routing is presented with simulation results which demonstrate that smart forwarding technique is effective in finding a low-cost path while it has the property of being able to find a feasible solution if there is one