This paper proposes RSS-based multiple-input multiple-output (MIMO) receiver diversity-combining techniques for identity-based attack detection. The diversity-combining techniques considered for identity-based attack detection using MIMO systems include selection combining (SLC), equal gain combining (EGC), and maximal ratio combining (MRC). First, we exploit the spatial correlation of the RSS hereditary from wireless nodes to detect identity-based attacks. Analytical expressions of the proposed method for the spatial correlation of the RSS hereditary using MIMO-combining diversities are provided and validated through simulation results. Second, we propose unsupervised machine learning (ML) algorithms such as k-means and k-medoids for the detection of identity-based attacks using RSS with different MIMO-combining diversities. Complete statistical derivations for the detection of identity-based attacks using k-means and k-medoids algorithms were performed. The simulation results confirm that the proposed techniques exploiting MIMO diversity-combining techniques for identity-based attack detection using unsupervised ML algorithms outperform the previously proposed techniques in terms of the false positive rate (FPR) and detection rate (DR). Furthermore, the simulation showed that the EGC-combining diversity with k-means and k-medoids outperformed the SLC- and MRC-combining techniques.