Research in Wireless Distributed Systems

This award is for the purchase of radio modems and portable computers for research into radio network (wireless) computer systems. The research will concentrate on several aspects of wireless communication including: extending the IP protocol to handle hosts that are not topologically fixed, techniques for handing off from one radio cell to another without data or communications loss, intelligent caching to reduce peak loads over slow radio links, self-organizing load balancing, and distribution of kernel functionality. Wireless networks provide an attractive alternative to the expense of running cables between and within buildings so long as high communication speeds are not required. This research is intended to solve some of the problems with wireless networks and to build an experimental network at Columbia University. It is very expensive to run wires or fiber optic cables between and within buildings. An alternative to cabling is to build a wireless computer network much like the cellular phone network is an alternative to the telephone line network. The problems with building a wireless computer network are more complex than those involved in building a wireless voice network. These problems include protocols to "hand off"communications when cell boundaries are crossed (with voice, some noise is acceptable, with data this must be minimized), load balancing methods so that slow radio links don't get clogged, and distribution of operating system functionality to reduce communications. The equipment funded by this grant will enable an experimental wireless computer network to be built. The network will be used to experimentally validate communication protocols and to validate algorithms for load balancing and other methods needed to reduce communications on the network.