System-Level Simulation and Resource Management for Distributed Antenna Systems with Cognitive Radio and Multi-Cell Cooperation with Imperfect Information
Ref: CISTER-TR-160101 Publication Date: Sep 2015
System-Level Simulation and Resource Management for Distributed Antenna Systems with Cognitive Radio and Multi-Cell Cooperation with Imperfect InformationRef: CISTER-TR-160101 Publication Date: Sep 2015
The performance of cellular networks will experience a considerable improvement by the use of distributed antenna systems (DASs), multi-cell cooperation (MCC), and cognitive radio (CR). However, several issues remain open in the system-level evaluation, radio resource management (RRM), and particularly in the design of billing/licensing schemes for these types of system. This paper proposes a system-level simulator (SLS) that will help us address these issues. An advanced RRM solution is also proposed for a multi-cell DAS with two levels of cooperation: inside the cell (intra-cell) to coordinate the transmission of distributed nodes within the cell, and between cells of a cluster (inter-cell) to adapt cell transmissions according to updated inter-cell interference measurements. The RRM solution blends network and financial metrics using the theory of multi-objective portfolio optimization. In this paper, each resource is considered as a financial asset whose allocation has to be optimized based on economic metrics such as return and risk (variation of the return). The core of the intra-cell RRM algorithm is based on an iterative weighted least squares (WLS) optimization process where power levels and beam-forming vectors are jointly designed to comply with a different SINR (signal-to-interference-plus-noise ratio) threshold for each transmission. This SINR threshold ensures the transmission of the selected modulation and coding scheme (MCS) with a target value of BLER (block error rate). The WLS scheme allows for a smooth integration of scheduling and adaptive modulation and coding with space division multiplexing. Convergence speed is improved by reusing the outcome of previous WLS iterations. The weights of the WLS optimization contain network (queue length and fairness), as well as economic metrics (return and risk). This process is complemented by a multi-objective portfolio optimization stage for joint spectrum selection and resource allocation that attempts to maximize return and reduce risk. Cells within a cluster exchange the results of their optimization processes to control inter-cell interference and so achieve MCC. All schemes use an imperfect copy of channel and queueing state information, which is the result of inaccurate measurements, imperfect feedback, or sensing errors.
Published in Recent Advances in Electrical & Electronic Engineering, Bentham Science, Volume 5, Issue 3.
Notes: Extension of the conference paper presented at FGCT 2015