A nonlinear detector derived within a maximum likelihood estimation framework is shown to be effective in retrieving the channel coefficients and data of users on the uplink channel of a noncooperative wireless system without the access point having any prior channel state information (no CSI or noncoherent setup). Rather than relying on pilot-assisted transmissions, it is shown that a maximum likelihood-based detector emerges naturally from an information-theoretic argument. The assumptions under which the detector is designed are as follows: 1) the uplink data from different users are independent and non-Gaussian; 2) the coherence block of the channel is much larger than the number of users (in practice, the square of the number of users); 3) the number of antennas at the access point or base station is equal to the number of users; 4) users continuously transmit within the coherence block; and 5) the transmission occurs at high signal-to-noise ratio. No coordination between the access point and unintended users (interference) is needed. Some coordination with intended users is needed. Finally, the system is assumed to be symbol-synchronous.