Spectrum sensing (SS) is one of the major advanced features of the cognitive radio (CR) technology. Although multiple SS approaches have been introduced in the literature, the spectrum sensing is still a challenging operation from the practice perspective. Therefore, in this letter, we go one step further toward proving the efficiency of this CR functionality within a testbed that aims to emulate a near real-world high-mobile wireless communication system, using the national instrument (NI) 2954R universal software radio peripheral as a software-defined radio platform. Indeed, SS existing experimental works consider only the time-invariant channel assumption. Yet, from a practical standpoint, this assumption is no longer viable for high-mobile wireless environments for military or high-speed railway applications, where the channel nature is rather time-selective. Thus, this letter provides experiments of the SS function under time-selective multiple-input-multiple-output channels. The suggested prototype was designed with respect to saving time and development efforts. Furthermore, we show that the proper selection of the SS algorithm can enhance the communication system performances in terms of increasing receiver velocity (v_r) and reducing sensing time while improving detection performances (P_d). For instance, we have found that the receiver can move with a five times higher v_r value while ameliorating its detection capacity by 20%. In addition, at a fixed v_r value, the sensing time can be 100 times shorter while achieving up to 21% of P_d increase.