Distributed load-sensitive routing for computationally-constrained flows

A network that provides not only connectivity but also computational resources to application flows will enable a new array of network services. For example, applications that require content adaptation can be deployed more easily in a network that provides integrated communication and computational resources. In this paper, we study the problem of finding a path for a flow that has both computational and bandwidth constraints. We present a distributed load-sensitive routing algorithm that generates precomputed routing information and optimizes the routing decisions for both applications and computational and communication resource providers. We show through simulations that our distributed approach performs comparably to a centralized algorithm and is more resilient to longer routing update intervals.