Robust multilayer design of wireless networks for massively distributed systems

Andrea Goldsmith
Stanford University

Future wireless networks must provide a robust communications backbone for massively distributed systems. However, most distributed system designs have little tolerance for communication faults, as we will illustrate with a distributed control example. Therefore, much research is needed in designing robust wireless networks that meet the application requirements of massively distributed systems. In this talk we discuss three of the main research challenges in this area. We first discuss fundamental capacity limits of wireless networks. These limits not only provide bounds on the maximum rates that can be exchanged within wireless networks, but they also lend insight into optimal network designs. Next we focus on dynamic resource allocation strategies for wireless networks. In particular, we show that dynamic resource allocation is essential when the network must meet strict delay constraints (as in control applications), or when energy in nodes of the network is limited. Finally, we generalize these concepts into a new multilayer network design paradigm. This paradigm incorporates adaptivity and information exchange across layers as well as robustness and diversity at each layer to compensate for unpredictable network variations. We will illustrate this design philosophy on a case study of distributed control over a wireless network.

Presentation (PowerPoint File)

Back to Massively Distributed Self-Organizing Networks