Abstract
Dynamic Spectrum Management (DSM) is an effective method for reducing the effect of interference in both wireless and wireline communication systems. This chapter discusses various DSM algorithms, including Optimal Spectrum Balancing (OSB), Iterative Spectrum Balancing (ISB), Iterative Water-Filling (IWF), Selective Iterative Water-filling (SIW), Successive Convex Approximation for Low complExity (SCALE), the Difference of Convex functions Algorithm (DCA), Distributed Spectrum Balancing (DSB), Autonomous Spectrum Balancing (ASB), and Constant Offset ASB using Multiple Reference Users (ASB-MRU). They are compared in terms of their performance (achievable data-rate) by extensive simulation results and their computational complexity.
TopBackground
This section provides some background material on the topic of dynamic spectrum management. For more detailed background material, interested readers are referred to Leaves et al. (2004), Akyildiz, Lee, Vuran, & Mohanty (2006), Zhao & Sadler (2007), and the references therein. A Venn diagram illustrating the similarity and differences between the algorithms discussed in the background material is shown in Figure 1.
Figure 1. Venn diagram illustrating the similarity and differences between the algorithms discussed in the background material
Yu, Ginis, & Cioffi (2002) introduced Iterative Water Filling (IWF), one of the first distributed DSM algorithms. In IWF, each user selfishly performs their own power allocation without any regard for their effect on the other users until a point where no user benefits from changing its current power allocation is reached. While IWF gives significant performance improvements over SSM techniques, in many situations, it leads to sub-optimal performance.
Key Terms in this Chapter
Centralized Algorithms: Algorithms which make use of a central hub with full knowledge of the network to perform the power allocations.
Sequential Convex Programming: Iterative algorithms which solve non-convex optimization problems by solving a sequence of convex sub-problems to find a locally optimal solution.
Semi-Centralized Algorithms: Algorithms which make use of both a central hub and distributed processing to perform the power allocations using a per-iteration message passing system.
Locally Optimality: Refers to an operating point which is the best within some neighbourhood (i.e., subset of the domain) with respect to some objective (e.g., maximum sum-rate, minimum transmit power).
Nash Equilibrium: A non-cooperative game theoretic solution where no user can gain by changing their current power allocation strategy.
Distributed Algorithms: Algorithms in which users self-optimize fully autonomously without the need of explicit message passing.
Dynamic Spectrum Management: Adaptively applies different spectrum allocations for each user with some optimization criteria in mind (e.g., maximizing throughput, minimizing power).
Global Optimality: Refers to an operating point which is the best possible over the entire domain with respect to some objective (e.g., maximum sum-rate, minimum transmit power).