In legacy DOCSIS, data can only be transmitted in one direction across any part of the spectrum.
Compared to the passive optical networks (PONs), a cable access network is severely limited in the maximum symmetrical data speed due to the upstream RF spectrum scarcity. Since bringing fiber to the home is extremely expensive, cable operators have searched for an alternative to deliver the multi-gigabit services promised. This need together with recent trends in the cable industry (i.e. the deployment with DOCSIS 3.1 Orthogonal Frequency Division Multiplexing (OFDM); the deep fiber migration; and the remote PHY network architecture) has resulted in the rapid development and standardization of the full duplex (FDX) DOCSIS technology. With FDX DOCSIS, the RF spectrum can be used simultaneously in both the upstream (US) and downstream (DS) directions, allowing up to 5 Gbps US service and 10 GbpsDS service over the cable access network.
In FDX communications, a system supports simultaneous bi-directional transmissions across the same spectrum. Interferences between the bi-directional transmissions therefore must be mitigated for the intended signals to be properly received. DOCSIS is a point to multi-point system, where multiple cable modems (CMs) are connected to the same Cable Modem Termination System (CMTS) port via a coax distribution line. When one CM transmits upstream to the CMTS, the US signal may leak through the cable plant and becomes interference in the DS direction at the receiving CMs. Since the source of the interference is unknown to the receiving CM, PHY layer echo cancellation cannot be used. FDX DOCSIS address this issue by grouping CMs that interfere with each other into an Interference Group (IG). CMs in the same IG must transmit or receive along the same direction at any given frequency and time. CMs from different IGs have enough RF isolations to allow simultaneous US and DS transmissions at the same frequency.
In this paper, we will discuss IG discovery, a new process introduced in FDX DOCSIS to determine the IGs based on the CM to CM interference measurement obtained via sounding. We will start by introducing the basic IG concept and the operational principles to conduct sounding. We will examine the system overhead in terms of the spectrum cost and the time to converge for sounding among a given number of CMs at the desired frequency granularity. We will then propose a set of optimization techniques to improve sounding efficiency. We further extend the solution space by incorporating aniterative IG Discovery model to allow the system to automatically adapt to the changing network environment for optimized system performance.