This paper integrates traffic engineering (TE) in network planning (network improvement) to build high performance networks, which achieve traffic’s multi-constrained quality of service (QoS). It is a NP complete problem that cannot be efficiently solved by traditional network improvement with extending equipments’ capability. A new network improvement algorithm based on TE is proposed. A heuristic algorithm of graph’s connectivity augmentation is presented to satisfy the topological constraint, a static routing algorithm based on multi-QoS requirements is adopted to satisfy TE constraints, and a genetic algorithm is used to globally search the network with minimum improvement cost and with its capacity of rational allocation. With the simulation analysis, while achieving network’s multi-constraint, rebuilding networks by the new network improvement algorithm is only a traffic balancing, but not a local blocking of the existing high performance networks.