Minimizing Distortion via Multiuser Resource Allocation

We investigate source–controlled resource allocation for orthogonal transmission of heterogenous sources over a multiuser system. The system resources bandwidth and power are shared among all users in such a way that the average distortion for all users is minimised. Especially in mobile communications, the allocation of resources among users with different applications is a challenging task. To illuminate this complex problem we consider simple models for source, channel, and distortion measure. For a constant channel rate the resource allocation reduces to a statistical multiplexing problem. In this case, only the source characteristics are exploited to optimally assign rate to each user. Additionally, if the channel noise is a random variable, either bandwidth or power levels can be assigned in an optimised way. For all three cases, the optimum schedule of resource allocation is derived including which sources are to be encoded and transmitted. Additionally, performance bounds on the distortion are derived for systems with large number of users. Fixed resource assignment, optimised bandwidth allocation and optimised power control are compared for a specific distribution of source variances and channel noise. Significant gains of optimised resource allocation can can observed whereby the bandwidth allocation is superior to power allocation.