Mandarim: uma solução para negociação de serviços com QoS em redes em malha sem fio utilizando agentes autônomos

Abstract

Wireless Mesh Networks (WMN) are composed by routing and client nodes, where these routing nodes have little or no mobility and form a wireless backbone. Such networks provides access to wired or wireless devices and its backbone integrates the WMNs with other networks such as the Internet, cellular networks, IEEE 802.11, IEEE 802.15, IEEE 802.16, etc. WSNs were projected to solve the limitations and to significatively enhance the performance of ad hoc networks and wireless metropolitan area networks (WMANs). These networks inspire a great number of deployments and will provide wireless services in different environment such as commercial establishments, university campus, community networks, wireless home networking among others. Besides recent advances in WMNs technology, there are still research opportunities in this area. Quality of Service (QoS), especially in multimedia wireless services, for example, is a crucial problem that has not being addressed by the literature. Although not trivial, provide QoS in wireless services is an important task for the proper operation of the applications that use the WMN to route its data. The identification os thenetwork requirements for the traffic created by the application is necessary to QoS provision. This dissertation proposes a solution, called Mandarim, to service negotiation in wireless mesh networks using autonomous agents with the objective of increase QoS for the servicesprovided by the network. Each network node has an autonomous agent responsible to service negotiation and are capable of deciding which nodes will consume or provide the negotiated service. Results obtained from a simulated environment shows that Mandarim increase the WMNperformance and QoS of the services provided the WMN nodes when compared to an architecture of best effort and the Resource ReSerVation Protocol (RSVP) protocol. The network throughput using Mandarin was, on average, 42% better than RSVP and 228% when using the better effort architecture. There were a 36.75% decrease in energy consumption and a 63% decrease in the number of TCP packages retransmitted. Mandarim also reduced packages lost by 68.5% when compared to RSVP and 41% against the best effort architecture. When consideringnetwork delay, reductions where, respectively, 96% and 26%.