RIA: Resource Control in Communication Systems with Time- varying Characteristics

这个计算机网络项目涉及控制 在大型通信系统中使用处理资源。 与 用户需求的数量和复杂性不断增加，因此需要 建立有效的资源分配机制。 重点 在这些系统中，基础驱动参数随 时间 该方法解决了当前几种方法中的两个缺点， 优化资源控制的方法： 假设某些参数由已知的 概率分布，驱动参数是 随时间静止或不变。 这种方法使用随机 在系统中在线执行的优化算法，其中 驱动参数随时间变化。 的收敛性质 使用相对较新的大偏差理论的算法 的研究，并通过模拟进行实验，允许 不同算法的比较。 预计该 结果将有助于通信资源控制 系统以及增加对时变 一般的随机系统。

This project in computer networking concerns controlling the use of processing resources in large communication systems. With user demands growing in number and complexity, there is a need to develop mechanisms for efficient resource allocation. The focus is on systems in which the underlying driving parameters vary with time. The approach addresses two shortcomings in several current approaches to the optimization of resource control: the assumptions that certain parameters are governed by known probability distributions, and that the driving parameters are stationary or constant with time. This approach uses stochastic optimization algorithms that execute on-line in a system in which the driving parameters vary with time. Convergence properties of the algorithms using the relatively new theory of large deviations are studied, and experiments conducted through simulation allow comparison of different algorithms. It is expected that the results will contribute to resource control in communication systems as well as to an increased understanding of time-varying stochastic systems in general.