Distributed resource allocation schemes
IEEE Signal Processing Magazine
Achieving high spectral efficiencies in wireless networks requires the ability to mitigate and manage the associated interference. This becomes especially important in networks where many transmitters and receivers are randomly placed, so that in the absence of coordination a particular receiver is likely to encounter significant interference from a neighboring transmitter. A challenge is then to provide a means for coordination, which allocates available resources, or Degrees of Freedom
... s of Freedom (DoFs), at each transmitter to avoid interference wherever possible. In a wireless network DoFs generally refer to noninterfering modes of transmission, and can be defined in frequency, space, and time. DoFs in frequency and time typically correspond to multiple nonoverlapping channels and time slots, respectively, and DoFs in space correspond to orthogonal spatial beams. Resources then include available power and beams at each transmitter, and the objective of the network operator is to allocate those resources among available DoFs to optimize an overall network objective. In principle, optimal resource allocation in a wireless network can be achieved if all active nodes measure all channel gains to all other nodes, and pass that information to a resource manager. After collecting this information the resource manager would then determine allocations over all users, or transmitterreceiver pairs, and announce those to the various transmitters. Of course, such a centralized scheme for resource allocation requires excessive information exchange and overhead for most practical networks. In addition, depending on the objective and specific resource constraints, the centralized optimization problem can be non-convex with associated worst-case complexity that increases exponentially with the number of users and DoFs.