Constructing covering codes via noising

We show how a combinatorial optimization method, the noising method, can be used for constructing covering codes. The noising method is based on descent. Starting with an initial solution, repeat the following steps: add noise to the data (in order to change the values of the evaluation function); and apply the descent method to the current solution for the noise data. For each iteration, the amount of noise is decreased until it reaches 0 in the last iteration. The final solution is the best solution computed during the process