1. Introduction
Wireless Ad Hoc Networks is a special kind of wireless communication network, which is a multi-hop autonomous system composed by a plurality of nodes with a wireless transceiver device. The node in the network has the function as both host and router [1]. Over the past decade, MANET is increasingly becoming a hot research field in computer networking and wireless communications [2]. Since MANET doesn’t require fixed infrastructure and has a strong invulnerability, it has a high application value. The main characteristics of MANET include sharing radio channel, high bit error rate, multi-hop communications and dynamic topology. In this context, it is a great challenge to provide QoS support [3]. On the other hand, although the MANET protocol stack has the similarity with the Internet protocol stack, the related protocols cannot be directly copied from Internet protocol stack due to the particular features of MANET [4].
QoS support in MAC layer is a crucial and challenging topic in the research of wireless ad hoc networks [5]. MAC protocol has the inherent advantage for QoS. The wireless network requires strict synchronization and is complex in the case of multi-hop mobile allocation and scheduling between nodes. If the network has a larger number of nodes, the complexity will increase sharply, which limits the application of mobility protocols in wireless ad hoc networks. Xiao, Y. et al. presented a two-level protection and guarantee mechanism for voice and video traffic in the EDCA-based distributed wireless LANs, which can support multimedia applications such as voice and video over the wireless medium, a contention based channel access function [6]. Ahn, G. S. et al. proposed a stateless network model, called SWAN, which uses distributed control algorithms to deliver service differentiation in mobile wireless ad hoc networks [7].
2. A QoS Model Design in MAC Layer
The 802.11e MAC layer protocol mechanism can be well differentiated services with strong ability to control the business of low priority. However, this method has abandoned the original AIMD algorithm rate control and shaping mechanism. If the best-effort service requirements of the application layer initiates too large, there will be a large number of low priority packets at the MAC layer queuing. These low-priority data transmission rate will be low with the heavier routing load. The low-priority queue is lined after entering the packet is discarded in the MAC layer. Application layer mechanisms can guarantee the retransmission of these packets discarded, but it can also result in additional network overhead. Therefore, the need to control the applicationlayer contracting the queue is full in the MAC layer. Therefore, a feedback mechanism is design. When low priority data queue is full in the MAC layer, the admission control module will detect this situation. The accepted admission control module can control the realtime data services and is responsible for the control of the business. When the MAC queue is detected, the system immediately stops sending best effort to lower business data. To avoid too frequent start and stop transmitting data to the lower layer, the MAC layer queue keep half empty when the admission control module to control data re-starts sending.
Figure 1 shows the model of the signaling frame structure, including probe frames, and initial re-negotiation frames. The frame structure is basically the same, and different frames can be distinguished from type domain. The type domain 0 means it replies detection frame to the destination. All copy detection frame is corresponding domain in addition to the type domain. The type domain 1 denotes the frame sender sends detection frame. It records the bottleneck bandwidth of the domain. The type domain 2/3 means this frame respectively for source-based or network-based re-negotiation message frame. The probe frames of bottleneck bandwidth with 16 bits are not used domain for recording.