Wireless Multimedia

Even before multimedia wireless, allocating the right amount of bandwidth had been a challenge. Allocate too little bandwidth in one part of the network, and wireless calls get dropped much more frequently when they are handed off from one wireless base station to another. Customer satisfaction can drop at the same time. Allocate too much bandwidth and carriers may not achieve the revenues needed to operate profitably. Now add to this more flexible wireless Internet Protocol (IP) networks — and data and video calls that accept different levels of service quality and successfully hand off with different levels of bandwidth — and the difficulty in predicting bandwidth needs increases dramatically.

In spite of the increased complexity, the new techniques developed by Telcordia researchers successfully predicted the levels of network resources that would both optimize the amount of traffic on the network and, at the same time, reduce the number of dropped handoff calls.

They did so because they model demand directly (unlike previous methods that try to model the myriad complicated and changing set of factors that affect demand). By modeling demand directly, the new techniques allow for wide variability in per-call resource demands, call and channel holding times, call arrival times, and network configurations.

The first technique used Wiener models similar to those that predict stochastic processes such as Brownian motion in physics or stock prices in finance. This technique used present network resource demands to predict future resource requirements.

Sometimes, future resource requirements may also depend on past resource demands. To take this into account, the researchers developed a second technique based on time series analysis.

Interestingly, these dynamic resource prediction techniques generated almost identical results that agreed well with the measurements of actual network resource demands about 95% of the time. The new techniques required fewer capabilities to run and were easier to implement compared to previous methods.

Leading the research effort was Prathima Agrawal. The team included Toshikazu Kodama, President of Toshiba America Research, which helped fund the research, Tao Zhang, Eric van den Berg, Jasmine Chennikara, and former Telcordia Research Scientist Jyh-Cheng. Their paper, "Local predictive resource reservation for handoff in multimedia wireless IP networks," was published in IEEE Journal on selected areas in communications.

In its annual competition, SAIC's Executive Science & Technology Council recognizes some of the most innovative research and best written technical papers and books by SAIC scientists and engineers. This article is a summary from one of the award winners.