"With the development and application of network technology, user mobility and reliability of the network increasing, the wireless Mesh network based on IEEE 802.11 family of standards has been rapid in recent years, it is widely used in wireless Mesh network, any the wireless device may simultaneously node as an access point (AP) and the router, each node in the network can send and receive signals, each node can directly communicate with one or more peer nodes. However, due to wireless network application characteristics itself and the various physical layer transmission techniques, appropriate media access control MAC protocol for wireless Mesh network is crucial.
MAC protocol applied in Wireless Mesh Networks comprises: CSMA / CA, DCF, PCF and the like, in order to achieve the MAC sublayer for supporting various QoS traffic flow, IEEE 802.11e working group IEEE 802.11 DCF mechanism is proposed based on the enhanced distributed channel access mechanism (EnhancedDistributed channel access, EDCA), wireless Mesh networks that can better provide audio and video services.
EDCA different traffic flow is divided into four different priority AC (Access Categories), each corresponding to an AC queue, by providing arbitration inter frame space (Arbitration Interframe Space, AIFS), the CWmin minimum contention window value, a maximum contention window and CWmax transmission opportunity TXOP (traNSmission opportunity) 4 parameter values to achieve service differentiation between different service flows. Literature studies have shown that, due to the static conditions provided a mobile wireless network and complexity, the EDCA algorithm does not make the four parameters to achieve optimal performance of wireless networks, high-load condition is large or bursty traffic in particular, As wireless networks have a higher collision rate, a sharp decline in network performance EDCA, the network can not meet user requirements. Also related on an adaptive control mechanism and related improvements backoff algorithm of CW through, such as Lamia Romdhani proposed AEDCF mechanism (Adaptive EDCF, AEDCF); rapid collision Younggoo Kwon proposed resolution mechanisms (Fast Collision Resolutio, FCR), making the EDCA algorithm is more suitable for wireless network environment. However, these studies did not consider itself EDCA algorithms and parameters AIFS, affect the CWmin, CWmax TXOP adjustment and fairness of the network wireless Mesh (between nodes and between different service flows) brings.
This paper proposes an algorithm based on fairness EDCA (Fairness-based EDCA, FEDCA). The basic idea of the algorithm is to determine the time FEDCA access category of data transmission and listens for this channel by a weighted round-robin fashion, the parameter is determined by calculating the TXOP fairness factor, under conditions to achieve fairness guarantees improve network performance and network the purpose of QoS guarantee. And the feasibility of the algorithm is verified by simulation results.
1 EDCA algorithm
Working Group IEEE 802.11e EDCA is the opportunity for QoS support made of the basic DCF access channel consistent manner, each mobile node CSMA / CA channel access mode is obtained by contention mechanism based on the IEEE 802.11 DCF protocol . At the same time provides more channel EDCA access category (AC) of different types of data transmission service, the service can be implemented to distinguish between different services.
About 1.1 EDCA algorithm
In order to ensure different QoS requirements of different services, the EDCA algorithm defines the upper class of service class 8 (Traffic Category, TC) layer 4, and this type of access categories (Access Category, AC) based on IEEE 802.1D, and TC are the class 8 AC class mapped to 4 queues: AC_VO, AC_VI, AC_BE and AC_BK, representing the voice (voice) class, video (video) class, best effort (Best Effort) class and background (background) class of business. In order to achieve four different priority queues difference AC, it defines four parameters: arbitration inter frame space AIFS, a minimum contention window value CWmin, maximum contention window CWmax and transmission opportunity TXOP. Different AC through different parameter settings, to control the process access channel, thereby realizing distinguish different traffic types.
A mobile node for data transmission in a queue by the two AC phases. In a first internal node contention transmission opportunity TXOP, obtaining transmission opportunity possible access to the queue channel access. Secondly, packet obtains a channel access opportunities then obtains a channel access opportunities through CSMA / CA in a manner different nodes can transmit data. EDCA algorithm to complete the first phase of data transmission tasks: different queues to obtain a transmission opportunity through competition.
The basic IEEE 802.11e EDCA access mechanism shown in Figure 1.
FIG 1 a basic access mechanism of IEEE 802.11e EDCA
When a collision occurs due to competition channel, then retreat into the process. In this process, the backoff counter Backoff Timer is set to [0, CW [AC]] any integer value in the range: Backoff_Timer (BT) = uniform [0, CW] × aSlotTime. CW [AC] The initial value of CWmin [AC]. When a collision occurs, the value of CW [AC] is increased to (CW [AC] +1) × 2-1, when CW [AC] is increased CWmax [AC], is maintained CWmax [AC] constant value , not increase. After successfully transmitting a data frame, the CW [AC] value is reset to CWmin [AC], and continues to sense the channel. A backoff timer to detect each idle slot, the value (BT) minus 1, the first data frame occupies the channel reduced to zero, if the plurality of nodes backoff timer while AC reduced to zero, the higher priority queue the data frame occupies the channel, other data frames has entered a new backoff procedure.
1.2 EDCA algorithm analysis
As can be seen from Figure 1, a higher-priority AC by providing a small AIFS, CWmin, and CWmax priority will gain access to the wireless channels to achieve different services of different service differentiation. IEEE 802.11e standard EDCA parameters given a set of recommended values for a web application in most cases. However, due to the different traffic flows in the wireless network itself mobility and scalability, dynamic changes in the larger network or the network traffic, the proposed value standard in Wireless Mesh Networks will each mobile node and a node unfair phenomenon, embodied in the following areas:
(1) AIFS, AIFSN setting value results in unfairness between nodes. IEEE 802.11e standard are given AIFS [AC] = aSIFSTime + AIFSN [AC] × aSlotTime. All mobile network nodes the AIFS, AIFSN same value, so it is possible to cause a phenomenon in plesiochronous network (a node in this competitive obtaining the channel so that the next competition to obtain the probability of a channel increases) appeared, resulting in a wireless network to other nodes repeatedly competition phenomenon can not get the channel frequently, so that different nodes access channel, shared resources unfairly, while further reduce the utilization of network links, quality of service affecting traffic flow.
(2) fixed AIFSN value results unfairness between different service flow. Due to the high priority AIFSN value of small, under more data in the high priority to be transmitted, the low priority traffic flows in competing channel still can not get the channel, will inevitably lead to lower priority traffic "hunger" phenomenon .
(3) CWmin and CWmax is set. From the point of view of the basic EDCA access mechanism, CW [AC] is the value of the key factors affecting data transmission and AC transmission data queue again after a failure to obtain competitive channel. CWmin and CWmax achieve business value, although the distinction between different services, but in high network load, it will also lead to low-priority services "hunger" phenomenon.
(4) TXOP setting. TXOP reflects the number of data frames to get the maximum transmission queue data transmission opportunity. If the reference value IEEE 802.11e standard will lead to unfair competition mechanism channels greater unfairness among the traffic flow.
(5) EDCA algorithm does not consider the mobility of the node and a channel interference caused influence of the error on the fairness of the network.
Based on this, to improve the QoS guarantees fairness, performance and network traffic flow of different wireless networks, FEDCA algorithm EDCA algorithm AIFSN, CWmin, CWmax, and TXOP parameter is adjusted based on four principles of fairness, in order to ensure between the mobile node and fairness between the different levels of service.
2 FEDCA algorithm
Based on the above analysis, a wireless network This section discusses in detail FEDCA algorithm implementation.
Algorithm to achieve 2.1 FEDCA
To ensure that the nodes between the mobile node and the same level of fairness different service flows, FEDCA algorithm process can be summarized as: weighted round robin scheduling, dynamic adjustment of the congestion window CW, and TXOP adjustment factor calculating fair.
(1) weighted round robin scheduling. FEDCA algorithm execution model as shown in FIG.
Figure 2 FEDCA algorithm execution model
Weighted round robin scheduling is thought to ensure fairness among traffic levels, a weight assigned to each sub-queue, data in different sub-queue scheduling in accordance with different weights, instead of using the minimum backoff algorithm EDCA queue window of access to data transmitted. The specific implementation process is the weight of each sub-queue is assigned a value corresponding to the AC W is [AC] (the weight value indicates the number of sub-queues can be continuously transmitted data), by polling data transmission mode for each sub-queue, if data in the queue is not enough to send a child once or Wi empty, go to the next sub-queue ready to send data, so take turns to perform.
Dynamically adjusting (2) the congestion window CW. To ensure among mobile nodes and mobile nodes within the same throughput and improve the fairness of the system of different levels of service, FECDA algorithm congestion window CW all business classes have adopted the first line of retreat exponential backoff a way that for any queue CW
CWmax, CWmax congestion window to maintain unchanged.
(3) fair and TXOP adjustment factor calculation. After each round of polling data transfer completion, the same mobile node to ensure fairness in the different levels of traffic flow, FEDCA algorithm for each sub-queue fairness factor F [AC] is calculated, and the fairness factor prespecified FD [ AC] comparison, by comparing the results for the next round of scheduling of each sub-queue size TXOP [AC] = (TXOP [AC] + ΔTXOP [AC]), which changes the specific relationship shown in FIG.
3 Fair factor F [AC] and ΔTXOP [AC] relationship map
2.2 FEDCA algorithm discussed
From FEDCA algorithm process point of view:
(1) fairness metric. FEDCA proportional fairness algorithm as a measure of fairness criteria, that is occupied by each type of service is proportional to the network resources, it can be implemented so that in addition to the outer fairness among traffic levels may improve the system throughput. FEDCA algorithm for each type of service distribution factor indicating a fair share of traffic class in the shared resource can be used in the mobile node; right (2) assigned to each sub-queue scheduling polling weighted value W [AC ] and the relationship between FD [AC]:
FEDCA algorithm
Queue data (3) FEDCA determination algorithm that can be sent by way of polling, consistent manner following a successful transmission which changes the congestion window CW EDCA algorithm, congestion avoidance after the failure of the transmission process is also consistent with the EDCA algorithm, but the congestion window FEDCA algorithm changes the method described in, the object is to maintain fairness for each level of service nodes.
(4) for each grade of service fairness factor FD [AC] is calculated as follows:
Where: Total-Length [AC] is the round robin scheduling queue is scheduled total length AC data; to ensure that each queue can be calculated in the current round of the queue scheduling fairness factor FD [AC], for a a queue should maintain a counter for the length of the queue statistics scheduling Total-length [AC].
(5) ΔTXOP [AC] is calculated as:
3, in order to reflect the distinction between different services ΔTXOP [high]> ΔTXOP [low]; F. [High] max> F [Low] max; F [high] min 
