This paper studies the use of meta-heuristic search algorithm Simulated Annealing for solving a minimum cost Network Synthesis Problem which is common in the design of telecommunication networks and considers a set of practical constraints including arc capacity, node capacity, node degree and hop limit. We propose an improved neighborhood structure based on dynamically building routes between O-D pairs. The advantage of this neighborhood structure is that precomputed traffic routes are unnecessary and the traffic flow can be dynamically allocated amongst these routes. The experimental results show the effectiveness and efficiency of the new algorithm and the ability of it to solve larger scale problems.