Many noncoherent distributed strategies for cooperative sensor networks that do not require channel knowledge at any antenna to overcome the overhead involved in channel estimation are lately suggested; however, these strategies suffer from low system performance in terms of bit error rate (BER) and a comparably high decoding complexity. Differential beamforming strategies have recently been proposed to overcome these problems; however, they are implemented using the four-phase protocol. Thus, we propose a new strategy based on the three-phase protocol to increase the symbol rate. By doing this, a significant improvement can be achieved in the overall system performance. Hence, in this article, a new bidirectional differential beamforming strategy is suggested: 1) to be applied on the three-phase protocol instead of the four-phase protocol; 2) to be applicable for a decentralized wireless sensor network using single-antenna sensors distributed randomly between the communicating base stations; 3) to enjoy low decoding complexity; and 4) to improve the network performance in terms of BER by maximizing the received signal-to-noise ratio at the receiving base station without requiring channel knowledge at any antenna in the whole network. From our simulation results, the proposed strategy shows a substantially improved BER performance compared with the current state-of-the-art ones.