OPNET Modeller is the network development software. OPNET permits design and learn communication networks, devices, protocols, and application (Prokkola, 2006). OPNET's object-oriented modelling stylishness and graphical user interface (GUI) let easy of developing models from the real world network, hardware devices, and protocols. Modeller supports all major network forms and technologies, permitting you to design and test various situations. (Opnet, 2009)
In this paper, various metrics for comparing the performance of Medium Access Control (MAC) protocols and a network model to carry out simulation is discussed. At the end of this chapter, results obtained from the simulation in the form of graphs will be presented.
Distributed Coordination Function, DCF
DCF is the fundamental MAC method used in 802.11 and is based on a CSMA/CA (Collision sense multiple access/collision avoidance) mechanism. A mobile station STA) is allowed to send after the medium is sensed idle or duration greater than a Distributed Inter-Frame Space (DIFS). If during anytime in between the medium is sensed busy, a backoff procedure should be invoked. Specifically, a random variable uniformly distributed between zero and a Contention Window (CW) value should be chosen to set a backoff Timer. (Dangjiang & Shen, 2003)
Enhanced Distributed Coordination Function, EDCF
The 802.11 MAC does not support the concept of differentiating frames with different priorities. The enhanced DCF is designed to provide differentiated and distributed channel accesses for frames with eight different priorities (0-7). (Sunghyun & Prado, 2008)
The metrics used are Throughput, Access Delay, and End to End Delay in case of real-time multimedia traffic like VoIP, Video streaming (Video conferencing), response time in case of Telnet or Remote Login; type applications which cannot tolerate delay and loss of data. Retransmission Attempts in case a station does not get a chance due to internal collision. The following list below is the item of metric's used:
The Throughput of all stations shows the utilization of the wireless medium. Wireless bandwidth was a scarce resource, so efficient use of it is vital.
Media Access Delay
Access delay measured as the time from when the data reaches the MAC layer until it is successfully transmitted out on the wireless medium. The reason for studying average access delay was that many real-time applications have a maximum tolerable delay, after which the data will be useless. Therefore, it is important to provide low delay for real-time flows.
Total number of Retransmission Attempts by all Wireless Local Area Network MACs in the network until either packet was successfully transmitted or it is discarded as a result of reaching short or long retry limit. For 802.11e-capable MACs, the Retransmission Attempt counts recorded under this statistic also include retry count increments due to internal collisions. This factor plays important role in Performance of WLAN.
Data Dropped due to unavailability of access to medium. This factor largely affects the reliability of WLAN.
In this simulation, the wireless topology consisted of several wireless stations and one base station in the wireless LAN. The base station was connected to a wired node (Figure 2.1) which serves as a sink for the flows from the wireless domain. All wireless stations were located such that every station was able to detect a transmission from any other station, and there is no mobility in the system. This means the results will not be impacted by mobility and phenomenon such as the hidden node problem.
Figure 2.1 Wi-Fi network models
The simulation experiments were carried out using OPNET Simulator version 9.1 on Windows XP SP3. For the simulation, a data rate of 11 Mbps was chosen. Various MAC and PHY (Physical Layer of OSI) parameter values used in the experiment were according to IEEE 802.11e default values given in Table 2.1 (Sharma, Ganesh, & Key, 2008). We have run the simulation for 5 minutes for each scenario, and then compared the results obtained from them. Figure 2.1 shows a network model for the experiment.
Table 2.1: MAC and PHY parameter values used in Experiment (Sharma, Ganesh, & Key, 2008)
Data Rate (bps)
Transmit Power (W)
Buffer Size (bits)
AP Beacon Interval (secs)
Large Packet Processing
To compare the performance of DCF and EDCF two scenarios were created; medium access in first scenario was supported by DCF and in second, EDCF protocol was used at the MAC layer. Network environment factors which were used as a benchmark configured same for both scenarios. Detailed specifications are given in the Table 2.1 showing the MAC and PHY parameters used in experiment. The performance evaluation is done by simulating both scenarios one by one in OPNET simulator and then comparing the graphs obtained.
After choosing metrics, the simulation is done for 5 minutes for a scenario. Then results were gathered.
Analysis of EDCF
In case of EDCF, all four traffic classes were fed into the MAC layer from higher layer, which are corresponding to AC (0), AC (1), AC (2) and AC (3) respectively to check how efficient the new protocol is to provide service differentiation required for real time application. (Note that DCF does not support service differentiation, so no provision of Access category). For this, in the application profile of scenario (for EDCF protocol) different application was configured for different access category. Details are shown in the Table 4.1.
Table 4.1 Access Category corresponding to an application
REMOTE LOGIN (HEAVY)
In the profile configuration, a profile for clients was configured that uses all the four applications. In simulation scenario, 15 stations were configured to use these services randomly. In the simulation, we assumed that each traffic class has the equal portion of the total data traffic in terms of the average number of packets generated per unit time. The results obtained are as follows:
Throughput of Different Access Categories
Figure 4.2 Throughputs of Different Access Categories
It is observed from figure 4.2 that the Throughput of Access category 3 is way high than the Access category 0 and 1. Throughput for Access category 2 lies in between 3 and 1. It means that Throughput for applications like Voice over IP and Video conferencing, EDCF provides maximum Throughput by providing them more priority over the other services like simple HTTP.
Media Access Delay for Different Access Categories
Figure 4.3 Wireless LAN - Media Access Delay
It is observed from figure 4.3 that the Media Access Delay for Access category 3 is at minimum among all Access categories. Media Access Delay for Access category 2 is just 3 to 4 seconds more than AC (3). It means that the medium is assigned to the application according to the priority. Thus, EDCF provides lesser Medium Access Delay for real-time applications.
Comparative Analysis of DCF and EDCF
Next step is to check the performance of both protocols in terms of Throughput, Media Access Delay, Retransmission Attempts and Data Dropped. These four metrics are determining factors in terms of overall performance of both the protocols.
Figure 4.4 Throughput of DCF vs. EDCF
It is observed from figure 4.4 that in the first 30 seconds of simulation, Throughput of both DCF and EDCF is high, but then after that, it decreases with time and stabilizes for both protocols. Throughput in first 30 seconds is high due to less number of Retransmission Attempts (less number of backoff's). From Graph analysis, one fact is clearly visible, that curve of DCF is marginally higher than that of EDCF. We can conclude that DCF's overall Throughput is somewhat more than the EDCF.
Figure 4.5 Retransmission Attempts of DCF vs. EDCF
It is observed from figure 4.5 that in the first 30 seconds of simulation, Retransmission Attempts for both DCF and EDCF are less, but then after that, it decreases with time and stabilizes for both protocols. Retransmission Attempts in first 30 seconds are less due to less number of backoff's assigned to wireless stations. There is a small noticeable difference between curves of Retransmission Attempts of DCF and EDCF protocol. That small difference implies that the overall Retransmission Attempts made in DCF protocols are a bit lesser than EDCF protocol.
Media Access Delay
Figure 4.6 Media Access Delay of DCF vs. EDCF
In Figure 4.6, for the first minute of simulation the Medium Access Delay for both protocols increases at equal pace, and then after that, DCF suffers somewhat lesser Access Delay than EDCF. The increase in the Medium Access Delay for both protocols is due to increase in the number of nodes competing to gain access of medium.
Figure 4.7 amount of Data Dropped of DCF vs. EDCF
It is observed from figure 4.7 that the first 30 seconds of simulation, DCF suffers a sudden high Data Drop, but Data Drop in EDCF increases gradually. The reason of varying Data Drop gradually in EDCF is the service differentiation which provides priority based scheme to handle different kind of data. After 2.5 minutes of simulation, curves of Data Dropped of DCF and EDCF remain same for both protocols, EDCF finishes at less Data Dropped than DCF.
The results gained from simulation shows that EDCF delivers efficient mechanism for service differentiation to the Wireless LAN. However, this enhancement comes at a cost of a decrease in quality of the lower priority traffic up to the point of starvation. The achievement of the radio channel by the higher priority traffic is much more aggressive than for the lower priority. Higher priority traffic benefited, while lower priority traffic suffered.
In terms of overall performance, DCF performs slightly well than EDCF. This occurs due to reason that in EDCF mechanism, each AC function acts like a virtual station for medium access, so more collision will be expected for EDCF scenario. But in terms of Quality of Service for real-time applications (like Video conferencing) EDCF out-performs DCF.
EDCF has been purposed as the medium access control protocol for IEEE's approaching standard IEEE 802.11e. Currently, all of the wireless devices use DCF as the default MAC protocol and PCF as the optional functionality.