Advanced Networking and Distributed Systems (ANDS) Status Report

Principal Investigator: Leonard Kleinrock

• We are examining how the trend of increasing data rates will affect the communications protocols of the future. We define the latency/bandwidth tradeoff which allows us to identify applications which can benefit from increased bandwidth and those that can't. This also contributes to developing flow control, routing, and storage algorithms for gigabit networks.
  
• We have developed algorithms and corresponding data structures which are computationally efficient in the evaluation of Generalized Stochastic Petri Net models.
  
• We have analyzed and simulated the behavior of a group of finite state automata, namely Goores, which exhibit very robust, adaptive, fault tolerant collective behavior. We have already designed a number of communications protocols for the Goores, and are extending this work to various other applications.
  
• We solve the task of determining when to balance the load shared among a group of SIMD processors. This research yields an equation which can be easily computed on the fly, and will automatically adapt to different size problems and to changes during the the course of a single problem.
  
• We have investigated the use of Genetic Algorithms in the area of modeling a data source. We find that a relatively small sample of data is needed by the GA to produce an effective model. These models can then be used to greatly speed up the testing and development many different telecommunications protocols.
  
• We have studied the performance of an input-buffered ATM switch. The goal is to alleviate the Head-Of-Line (HOL) problem inherent in input-queueing, FCFS schemes. Under various traffic patterns, our model significantly increases the switch's throughput.