This paper presents a generalised spatial modulation (GSM) with lattice reduction (LR) aided K-best decoder for multiple-input multiple-output (MIMO) systems, achieving near optimal performance with low complexity. GSM is one of the current feasible solutions alleviating the requirement of high number of transmit RF chains in large scale MIMO systems. It conveys information by activating a subset of transmit antennas to reduce the transmit power and design complexity. In our proposed system, either the same or multiple information bits can be transmitted through multiple antennas achieving diversity gain and spatial multiplexing (SMx) respectively. In addition, as a MIMO decoder at the receiver side, a LR-aided K-best decoder for both real and complex domain is incorporated in order to obtain near optimal performance with less complexity, compared to a maximum likelihood (ML) decoder. Following IEEE 802.11 standard, we develop the decoder for 4x4 MIMO for different modulation schemes, with 2 active antennas at the transmitter side. The simulation results show comparable bit error rate (BER) performance between GSM with ML and the proposed scheme using both SMx and diversity gain. However, GSM with SMx utilises lower modulation order to achieve same spectral efficiency and thereby reduces the computational complexity.