Switching under energy constraints

The performance metrics that network designers typically optimize their design for are: throughput, delay and queue back-logs. These metrics have guided the design of wireline networks, like the Internet, and network components (e.g. switches, routers, etc). In wireless networks, the energy consumed by transmission nodes is a precious resource whose conservation is very important. This has motivated the research community to design wireless networks so as to deliver high throughput, achieve low delays, and minimize energy consumption. This paper makes a start at understanding how minimizing communication energy impacts the design of high-performance network algorithms. Specially, we consider a "switch" topology as this captures most of the constraints of a general acyclic wireless network. We study the problem of designing minimum energy transmission schedules which deliver maximum throughput. We show that the optimal policy is the solution of a convex optimization problem. We provide an iterative scheme that exploits the structure of the problem to obtain the solution. We also obtain approximations which are implementationally simpler and interesting in their own right.