Large network average path length will cause large network delay which brings difficulty in supporting the time sensitive services and applications. Large hop distance between source node and destination node in traditional network leads to significant network delay. By adding long-ranged links, path length from source node to destiny node will be reduced and original network can be transformed into a scale-free network with a small network average path length. The network delay is optimized by minimizing hop distance, in which information can transfer more efficiently and rapidly. Adding links can lower network delay effectively, but on the other hand, it will increase its cost. Common network construction methods focus on separating networks that are very different from each other and mostly unaware of each other, such as fixed and mobile networks planning. But in many real networks, networks are dependent on each other; therefore ignoring these network interactions cannot become more efficient. Cost and effectiveness play a key role in real network construction and layering network is an effective way to analyze coupling network especially in heterogeneous network. In this paper, the model of a toward cost-effective scale-free coupling network construction method is proposed. It combines the advantages of layered network and cost-effective network. A layered coupling network model is established in which network is divided into several networks based on link property. Links in the same layer have the same property and the upper layer capability is higher than lower layer capability. The nodes in the upper network are selected from the lower layer network coupling with the corresponding nodes with the same spatial location. Based on the network optimization and evolving network researches, the increases of node degree and local network radius are supposed to be continuous, moreover cost-effective indicator is introduced which characterizes the costs and effectiveness of adding links. Based on continuum, links are added to upper layer network with a certain probability by two continuous processes and thus network evolves into a scale-free network. The two continuous processes include node degree increasing process and local network radius increasing process. In the previous processes, cost-effective indicator is introduced and only the links satisfied cost-effectiveness are added. Cost-effective indicator characterizes the cost and effectiveness of network construction. Cost is proportional to Euclidean distance and effectiveness includes revenue of network average path length decreasing and link property benefit. In the coupling network, traffic prefers to transmit in the upper layer network for reducing network latency, and consequently leading to traffic congestion in upper layer. In the simulation, network topology evolution and dynamic traffic performance are evaluated. The simulation result shows that this method can effectively reduce the network latency within cost-effective requirement and initial network characteristics are maintained. The results also show that the network average path length declines slowly when network average path length is small because lower average path length needs higher cost when average path length is small. To investigate the traffic behaviors in the coupled layered networks, the traffic dynamic transition model is taken and dynamic traffic performance is given in this evolved scale-free network. Moreover, the cooperation between the two layers can be used to optimize network traffic performance by adjusting the link capacity to satisfy the requirements for the network congestion.