Dalarna University's logo and link to the university's website

In the wireless sensor networks, the hardware limitations of sensor nodes cause high transmission failure rate. We usually increase the density of nodes to improve the quality of information transmission. However, it is difficult for the limited energy supply, storage, and communication bandwidth to transfer large amount of redundant sensory data. So we use data fusion technology to remove the redundant data as much as possible before the data transmission. Data fusion becomes a research hotspot in recent years. In this paper we propose a multilayer and multi-agent data fusion mode, and analyze the proposed mode performance in three aspects: hops, energy consumption and network delay. The simulation experiments show that, if reasonably suitable parameters, such as the network scale, the number and size of agents, the data processing cost, are selected, the mobile agent mode is much better than the client/server mode.