In this paper, we evaluate the effectiveness of compressed-sensing-based channel estimation in LiFi orthogonal frequency-division multiplex (OFDM) systems with multiple-input multiple-output (MIMO). The sparseness of LiFi channels suggests that it is beneficial to utilize compressed sensing which has been investigated also for radio communications. In this paper, we formulate the mathematical problem of compressed-sensing for feedback generation in OFDM-based LiFi systems. We use well-known algorithms for both, sparse pilot design and sparse recovery and apply them to LiFi OFDM channels. We evaluate the effectiveness of these algorithms by measuring the achievable mean squared error (MSE) and vary the total numbers of subcarriers as well as the number of pilot subcarriers.