A Novel Approach for Profile Optimization in DOCSIS 3.1 Networks Exploiting Traffic Information

Abstract

The latest release of the data over cable service interface specification (DOCSIS), namely DOCSIS 3.1, has introduced the use of adaptive bit loading profiles. The design of these profiles is critical for the overall performance of a hybrid fiber coaxial (HFC) cable system. Previous works on profile optimization for DOCSIS networks have mainly focused on grouping cable modems (CMs) based on their signal-to-noise ratio (SNR) similarity to maximize the system’s capacity. However, this approach can be inefficient due to effect of data demand and traffic variations that can highly affect the overall system performance. In this paper, we introduce a profile design approach to maximize the average throughput of an HFC system while considering the effect of both CMs’ traffic and channels SNRs on the system’s performance. Moreover, the profile optimization should not require knowledge of future instantaneous data arrival rates, which is not possible to obtain in practice. Thus, our proposed approach uses only average information which can be easily predicted using learning techniques. We show that the proposed approach yields significantly high performance, particularly for scenarios with high degree of heterogeneity in the CMs’ traffic.