Several applications in a large scale network depend on diffusion of information from one node to others. In this paper, we propose a diffusion protocol for networked multi-agent systems based on both the structure of the network and the priority of agents. The agents (nodes in the network) interact with their neighbors for the diffusion of information based on the weighted difference of the resources available at the neighboring nodes. We observe that the system finally reaches to a weighted agreement which is proportional to the priority of the nodes, or the degree of the nodes. We also perform a convergence analysis of the proposed protocol on both static networks as well as dynamic networks. We propose three different applications where such diffusion protocols can be used – (i) in the detection of link failure, (ii) for ensuring security and utilization of network resources, and (iii) to come up with a static fixed point convergence over a dynamic network. The impact of stability and convergence of the proposed diffusion protocol have also been analyzed through simulation results under different test scenarios. The observations from the simulation validate the theoretical findings of the diffusion protocol and its applicability under different applications.

中文翻译：

---

多代理系统中链路故障检测和资源利用的扩散协议

大规模网络中的若干应用依赖于信息从一个节点到其他节点的扩散。在本文中，我们提出了一种基于网络结构和代理优先级的网络多代理系统扩散协议。代理（网络中的节点）基于相邻节点可用资源的加权差异与其邻居交互以传播信息。我们观察到系统最终达成了一个加权协议，该协议与节点的优先级或节点的程度成正比。我们还在静态网络和动态网络上对所提出的协议进行了收敛分析。我们提出了三种不同的应用，其中可以使用这种扩散协议——（i）在链路故障检测中，(ii) 确保网络资源的安全性和利用率，以及 (iii) 提出动态网络上的静态定点收敛。还通过不同测试场景下的仿真结果分析了所提出的扩散协议的稳定性和收敛性的影响。模拟的观察结果验证了扩散协议的理论发现及其在不同应用下的适用性。