In this paper, interference management for cognitive radio (CR)-enabled smart grid communication is investigated. Here multiple smart meters (SMs) communicate with the control centre simultaneously through a data concentrator unit (DCU) while co-existing with the primary user of the CR network and thus, sharing the same resources. The impact of the resulting interference between primary transmitters (PTs) and SMs is studied through the mean square error (MSE) formulations at the DCU, control centre and primary receiver (PR). Linear precoding techniques are then designed to minimize these MSEs while keeping the interferences within a pre-defined level. Moreover, the error performance of CR-enabled smart grid communication is investigated for a benchmark low voltage (LV) feeder, adopted by CIGRE (Conseil International des Grands Rseaux lectriques). The results indicate that when the PT moves away from the PR (and the DCU), the error performance at the PR degrades while the error performance at the DCU and the control centre improves. In all the cases, optimum precoders and decoders have been found to improve the error performances at the DCU, control centre and the PR.