Blind signal detection by basic ant colony optimization algorithm which is limited to slow convergence speed and local optimum. An improved ant colony optimization algorithm of the direct blind signal detection is proposed in this paper. The algorithm adjusts the pheromone update methods of ant colony algorithm, adds the corresponding parameter control in the local update rule and the global update rule of pheromone. Simulation results show that the improved algorithm has fast convergence speed and stable performance compared with the literature algorithms.