Quantitative Dynamic Interdependency Measure and Significance Analysis for Cross-Layer Design under Uncertainty

Interdependency among system parameters may significantly affect the performance metric of interest for cross-layer design and optimization. However, it is very difficult to derive the interdependency among system parameters in a dynamic complex networking system due to uncertainty of data observation and system modeling. Furthermore, current cross-layer design normally includes many system parameters, making the multidimensional cross-layer optimization problem very difficult to solve. In this research, we proposed a new approach for dynamic interdependency measure and significance analysis in cross-layer design and optimization. The major contributions made in this paper are: (1) interdependency among system parameters under uncertainties is measured by using non-additive measure theory; (2) quantitative significance analysis is proposed for dynamically identifying what system parameters have the most significant effect on the performance metrics of interest for system objective function; and (3) we develop a new fast multidimensional optimization method for cross-layer design based on dynamic interdependency measure and significance analysis. We show the effectiveness and feasibility of the proposed approach for cross-layer design and optimization in IEEE 802.11 WLANs.