Aiming at fast hardware processing, a detection algorithm for multiple-input multiple-output (MIMO) systems that employ spatial modulation (SM), or SM-MIMO, is proposed. Conventional SM-MIMO detection inherently suffers from the nearsightedness as it depends only on the cost of the preceding path that has been traversed. To overcome such a drawback, the proposed detection traverses the tree by taking into account not only the cost of the preceding path but also the estimated cost of the sub-tree that has not been visited yet. Compared to conventional methods, as a result, the proposed algorithm significantly reduces the number of visits that determines the latency in hardware, making it suitable to hardware implementation