Autonomous Spectrum Management (ASB) for Asynchronous DSL

Abstract — Dynamic spectrum management is an area of grow-ing importance in DSL research and can dramatically increase data rate and service reach. Unfortunately most DSM algorithms operate under the assumption of perfect synchronization between the DSL modems within a network, an assumption that is almost never satisfied in practice. Lack of synchronization means that modems see crosstalk from one-another not only on the corresponding tone, but on neighboring tones as well. This inter-carrier interference increases the difficulty of optimizing the transmit spectra, coupling the spectrum management problem among tones. In this paper we propose two power allocation algorithms for the asynchronous DSL network, based on the Autonomous Spectrum Balancing (ASB) algorithm. We name them Asynchronous ASB (A-ASB) algorithms, which include Greedy Power Shuffle (GPS) algorithm and High SINR Approxi-mation (HSA) algorithm. Both A-ASB algorithms are completely autonomous and have lower complexity than the current state-of-the-art algorithms proposed in [1]. Convergence of the algorithms is shown under reasonable conditions that are satisfied in typical DSL channels. The GPS version of the A-ASB algorithm achieves significant better performance than the ASB algorithm. I