Joint Message Scheduling and Drop Policies for Many-to-Many Communication in Delay-Tolerant Networks

Many-to-Many (M2M) communication, which allows a group of nodes to communicate with each other simultaneously, has been shown to help battle contention for scarce resources in Delay Tolerant Networks (DTNs), and hence improve their performance. In this paper, we address for the first time the problem of joint message scheduling and drop in a DTN which employs M2M communication. We develop a generic framework, assuming the routing algorithm provides us with a utility function (derived in order to optimize some desired metric) which helps us compute utilities associated with each message. The goal of our joint message scheduling and drop framework is to maximize the increase in total utility. We formulate the problem of joint message scheduling and drop in DTNs that employ M2M communication as an integer optimization problem and show how to reduce it into a number of many-to-one problems. In order to solve the many-to-one integer optimization problems obtained, we develop an efficient but 1/2 -approximate greedy algorithm (Almost-Greedy), and use it to develop an optimal solution (DynOpt). Finally, we use DynOpt to develop an (1 - ∈)- approximate Fully Polynomial Time Approximation Scheme (eOpt) that lets us choose a tradeoffs between complexity and accuracy. The problem of joint message scheduling and drop in DTNs employing the traditional one-to-one communication is a special case of our problem where the communicating group size is limited to 2 and hence our contribution provides a theoretical formulation and theoretically proven solutions for that problem too. We compare our algorithms against adaptations of the state of the art one-to-one scheduling and drop policies to the M2M communication case using simulations and find that our algorithms significantly improve the performance of the network.