Resource Allocation in Cellular Communication Systems

这是麻省理工学院的Dimitri Bertsekas和加州理工学院的Andrea Goldsmith的联合项目。支持新兴多媒体无线应用的信道容量的稀缺性要求有效的策略来利用可用带宽。任何使用固定频率复用和基站分配的系统都必须根据最坏干扰假设进行设计。动态资源分配是一种更有效的策略，其中频率、基站、数据速率和功率水平是相对于当前和潜在的系统条件动态分配的。在这项工作中，我们建议将一般资源分配问题表述为受干扰约束的优化问题。代价函数很一般。它可以在优化框架中包含大多数实际系统复杂性，包括由于用户移动性、多类流量和信道变化而导致的小区切换。我们考虑了FDMA和CDMA两种划分可用系统带宽的方法。我们将应用线性规划来最小化瞬时成本函数，以及动态规划技术来最小化无限视界上的成本函数。动态问题的表述将利用神经动态规划技术的最新发展。

This is a joint project between Dimitri Bertsekas, MIT and Andrea Goldsmith, Caltech. The scarcity of channel capacity to support emerging multimedia wireless applications requires efficient strategies for using the available bandwidth. Any system using fixed frequency reuse and base station assignment must be designed relative to worst-case interference assumptions. Dynamic resource allocation is a more efficient strategy, where frequencies, base stations, data rates, and power levels are dynamically assigned relative to current and potential system conditions. In this work, we propose to formulate the general resource allocation problem as an optimization problem subject to interference constraints. The cost function is very general. It can incorporate most of the practical systems complexity in the optimization framework, including cell handoff due to user mobility, multiple classes of traffic, and channel variation. We consider both FDMA and CDMA methods of dividing up the available systems band width. We will apply linear programming to minimize the instantaneous cost function, and dynamic programming techniques to minimize the cost function over the infinite horizon. The dynamic problem formulation will utilize recent developments in neuro-dynamic programming techniques.