Skip to Main Content
Adaptive MIMO-OFDM systems employing eigenmode based signalling have a great potential to increase the spectral efficiency when the channel state information (CSI) is accurately known at transmitter (TX) side. However, the perfect CSI is a too strong assumption for a wireless system operating in frequency selective channels. In the presence of CSI errors, the eigenmodes orthogonality is lost and a spatial equalizer is used at each subcarrier to remove the inter-eigenmodes interference. In this paper we propose using a first order matrix inversion approximation (based on truncated Neumann expansion) to find an upper hound for the covariance matrix of the decision variable at equalizer's output. Based on this upper-bound we are able to find a new bit and power loading algorithm which maximizes the throughput subject to maximum transmit power and maximum frame error rate constraints. The effect of CSI errors on the achievable spectral efficiency is studied by computer simulations for different antenna correlation setups. The results clearly show that the proposed method is robust against CSI errors and channel spatial correlation. The achieved spectral efficiency at low and medium SNR is larger that the outage capacity with no CSI at TX side.