Access Point Selection Mechanism In Wireless Lan

Published Date: 02 Nov 2017

The performance analysis is the method to analyse the output from the possible simulations or experiments to come out with specific solutions for any problems. The analysis regarding the performance of Wireless Local Area Network (WLAN) or Wi-Fi is the study to evaluate the possible outcomes from the experiments that had been carried out in the terms of parameters such as number of Access Point (AP) and performance metrics such as throughputs. From the research paper that had been analyse, the APs are one of the most important factor to determine the performance of one specific wireless network. APs also determine the throughputs whether in one network have the maximum throughput utilize or vice versa. In maximizing the throughputs, the mechanism or algorithm had been proposed by most of the researchers. The researchers also come out with these varieties of methods to pick out the better method in order to maximize the performance in the conditions of increasing number of stations (STA), or clients or even any possible communications in a network. Thus the mechanisms to the best solution of maximization of throughput must be evaluated. The selection of APs are also one of the issues rises nowadays (Nicholson, Chawathe, Chen, Noble, & Wetherall, 2006). Not just that, the AP localization is also important to determine the selection of APs. This is because the position or location of AP will give impact such as noises and delay to the client that present in the network. The APs that are locate so close to each other will cause the interference and this will affect the performance of wireless LAN. The increase in time of end-to-end delay and longer response time will be the impacts of these interferences. Thus, the optimal solution to combat this problem is to determine the optimal mechanism of AP selection. These term papers will analyses and study the performance of the network from methods of algorithms or mechanism in the selection of APs that are proposed by researchers.

Literature review defined all the possible definition and explanations regarding this related information that are being research in this paper. Not to mention, literature review also review the related works, related theory and calculations that had been carried out.

2.1 Wireless Local Area Network (WLAN)

Wireless LAN is one type of communication which uses the medium of air as a channel to send and receive any type of data and information as according to Stallings (2001). This type of network usually acts as extension to the wired LAN in term of possible limitations such as distances and cost. Wireless LAN also have better features in which wired LAN are not supporting such as authentication and encryption. Not to mention, the use of wireless LAN can help reducing the cost of upgrading, maintenance and implementation. Wireless LAN consists of different architecture such as stations (STA), Basic service set (BSS), Extended Service Set (ESS) and also Distribution Set (DS). The station includes end devices such as receivers, and clients that are communicating with AP in a network. BSS is one type of architecture that that consists of only one transmitter or AP and numbers of client. While, ESS consists of more than one AP in a network in which that these APs are connected to another APs with the present of distribution system (DS). In other word, ESS is proposed as the BSS architecture is unable to support the network with a very large number of clients and a very big geographical area. In wireless LAN, the STA will have to connect to the AP to allow any other communication. In other words, for the STA to connect to other STA, the AP must have been the in between communication or channel to allow the communication. In one large geographical area, there perhaps exists more than one AP however, the large number of STAs presented might degraded the overall throughputs as many STAs will connect to the same AP that have the high signal strength. This indeed is one of the major issues that needed to be overcome nowadays for the better transmissions.

Performance Analysis regarding the Access Point Selection in Wireless Local Area Network.

Access point selection in Wireless LAN is a method of connectivity between the AP and station (STA) or how the STA will choose the any AP to connect to. There are approaches that are proposed by the researchers and developers. However, these different kinds of approaches are still uncertain and are still under the experimenting stage before they can be implemented in the concurrent network infrastructure. Currently, most of wireless LAN network are implementing the use of signal strength approach or Received Signal Strength Indication (RSSI) based mechanism on AP selection. In this approach, the STA will choose or connect to the AP that has the highest value of Received Signal Strength (RSS). This method is still being implemented until now however, recent research had indicated that this current technique had major issues in degrading the network performance. This is because; the large number of STA in one area will connect to only one AP that reside the most value of RSS. Thus, this will lead to performance degradation if too many clients trying to communicate with only one AP in a time. There are modifications made based on the RSS-based AP selection scheme. The network management system (NMS) is one of the modifications to alter the problem of RSS-based AP selection scheme. NMS consists of software and hardware working together as monitoring tools to monitor traffics and notify those traffics to the STAs in the network. Not to mention, the NMS also act as a controller to the load-balanced. However, the intervention of NMS tools sometimes congested the traffics and network with lots of unnecessary information. The approaches of AP selection consist of the current centralized mechanism and proposed decentralized mechanism that is being researched by most of contributors. The idea behind this paper is to investigate the performance regarding the concurrent algorithm of RSS-based AP selection with the optimal and suitable AP selection mechanism to be the better solution to be implemented if the decentralized approach is going to take over the centralized approach.

Most paper that is reviewed agreed that their proposed methods are being compared with the RSS-based AP selection scheme. According to Fukuda, Fujiwara, Tsuru and Oie (2005), the RSS-based AP selection scheme is based on the centralized approach. The idea of centralized approach is to have one controller system that can manage most of the bandwidth, signal and other aspect of parameter for the STAs to use in order to connect to the specific and balanced AP. In the other hand, this controller will manage the balanced communication between the AP and clients. While, the idea for the decentralized approach is to have the specific algorithm of choosing the AP within the STAs, or the STAs their self select the AP based on some algorithms. As reviewed, some of the algorithm that uses the decentralized approach use the Maximizing Local Throughput (MLT) algorithm. In this algorithm, STA will optimize its own throughput locally. This algorithm can be a benchmark for the researchers to proprosed much better algorithm. The MLT algorithm is being use in the paper of Fukuda et al. as compared to the work of Lu et al., they are using the knowledge model instead of the MLT algorithm. From their work, they stated that using 0-1 function to classifying user is more suitable.

From the paper of Xue, He, Zhu, He and Liu (2012), they study the AP selection based on the Sociality-awareness. They had stated that the sociality in selecting the AP is important as it can unbalance the AP and cause the major degradation in throughput. Based on their paper, they proposed the AP selection algorithm based on the load-balancing scheme. As compared to the paper of Fukuda, Fujiwara, Tsuru and Oie, they have the same similarities in which that they had mentioned that the RSS-based AP selection might involve lots of client and only one AP in a large geographical area. In this term, the AP is unbalanced and load balancing is indeed significant to balance up the AP for the better performance.

While, Lu and Wu (2007) also stated that AP selection problem is categorized as one of the load balance problem which similar to the statement of Xue et. al. They also mentioned that the RSS-based AP selection also lead to poor performance. Not to mention, Lu and Wu also stated that the AP selection problem includes the knowledge of wireless user regarding the present of AP and also the timing of the AP selection on the different users. This case would show that the AP selection method is based on first come first serve mechanism. The STA that already associate with the AP will not lose it connection until the transmission is completed or the AP had powered-down. This in the future will cause the connection problem to other STA that present within the geographical area that also wants to connect to the AP. Thus, Lu et. al had come up with the localized heuristics AP selection algorithm based on the different knowledge model. As compared to the Sociality-awareness type of AP, they have the similar fact that location-based and sociality-based are closely related. This is because, location of positioning of AP have the relevant connection with the kind of sociality that happened in the network. The sociality that occurred frequently will cause that part of geographical area on the network condensed as same as logical explanation for the position of one AP that is so closely located to so many clients in one large geographical area. The reason behind this that the positions of AP and client have to be located balanced through the area and the sociality of the connection also has to be filter and balanced out properly.

As compared to the work of Abusubaih, Gross, Wiethoelter, and Wolisz (2006), they also manage to proposed the new static AP selection algorithm and enhanced static AP selection algorithm that consider the factor of most parameter such as STA own throughput and also the throughput of other STAs. Not to mention, the authors also believe that the STA will associated with the high signal strength AP that most of the reviewed contributors had mentioned. The author had explained that RSS-based AP selection took place as an active scanning activities in which the STA actively probing for the AP present in the network to be connected to.

As most of the contributors reviewed in this paper had mentioned that the distributed mechanism are more preffered as an alternative method for RSS-based AP selection, the centralized mechanism that needed the use of Network Management System (NMS) can be significant as a controller in a network. The NMS that said to be more costing that distributed system, can also perform the monitoring of traffic loads of AP present and send the information to the STA presented in a network. However, the NMS can be unfavourable when it comes to network congestion (Papaoulakis & Patrikakis, 2008). The NMS indeed as centralized approach seems not being the preffered mechanism to any of the contributors reviewed.

On the other hand, Papaoulakis and Patrikakis (2008) had also agree that current RSS-based is not good because it did not even consider the loads of AP. The authors also proposed the selection mechanism that consider the signal level and also the loads of AP in order to balance out the traffic between existense Aps in the network. The proposed method is also based on the current method of RSS-based AP selection and with no intervertion of NMS.

Figure 1.0 : (a) The unbalaced congested network and (b) The balanced and uncongested network (Source by: Tüysüz & Mantar, 2010)

The figure (a) above shows the unbalanced network that is produced by the RSS-based AP selection. While, figure in (b) shows the balance network that is going to be achieved. The congested networks that will reduce the performance of the network reside from the emergence of NMS or network management system as being reviewed by Papaoulakis et. al. From this figure, it clearly shown that the centralized approach such as the use of NMS cannot be used. Thus, the decentralized approach are more significant. The reason is that the decentralized algorithm

In the paper by Tüysüz and Mantar (2010), they proposed the algorithm that based on quality of voice and other important aspects such as utilization. They also said that this proposed method is suitable in real-time and even non real-time traffics. The figure below shows the example of AP selection algorithm that must be consider in every distributed algorithm. The idea behind this to consider other performance metrices rather than only consider signal strength as the method to choose AP. As compared to the work of other method proposed, they utilize most of formula that depends solely on the surrounding factor of the network rather than the nature of transfer of information. They consider most of the unbalanced factor of the AP and not the quality of the information.

Figure 2.0 : The example of AP selection algorithm that consider all important performance parameter compared to the concurrent RSS-based AP selection algorithm. (Source by: Tüysüz & Mantar, 2010)

Four out of six authors have concluded that the AP selection problem is also load balance problem. Thus, the algorithm such that being demeonstrated by Tüysüz and Mantar (2010), are the most preferable algorithm that can be the benchmark to proposed another new algorithm.

Conclusion

Wireless LAN act as extension of the wired LAN and are indeed are being utilize in the most of the network widely. With additional features like mobility, wireless LAN can support infrastructure such as BSS and ESS. In the wireless LAN, AP or sometimes transmitter is a very important element in managing the communication as every STA that wish to communicate will communicating through the AP. However, as the number of STA increases, the throughputs in the network will degraded bit by bit. This will further degraded most of the performance of wireless network such that the delay will be increase and the response time will be longer. This will made that those network are not preferable to be used. The AP selection mechanism had proved to be an important aspect in the advancing of wireless LAN. The AP selection mechanism is divided into centralized and decentralized approaches. These two approaches indeed have the advantages and disadvantages in the maximizing network performance. We can compare that all of these contributors mentioned earlier had agreed that the RSS-based AP selection are not productive at all and degraded the performance as the STA will only choose the AP that have high signal strength without evaluate other important parameter such as the throughputs or traffics of other nearest STAs. There are various of algorithm that are being proposed to be an alternative algorithm after the RSS-based AP selection such as decentralized approach by both Fukuda et. al and Fujiwara et. al. Other than that, there is also authors that proposed MAC mechanism algorithm that sound promising because they demonstrated the network improvement of 120% rather than any other algorithm. However, most of the methods or algorithm did not give the actual results of the simulation or real-time evaluation toward the comparison they made with the concurrent algorithm. All other method that had been reviewed agree that their algorithm increase the performance factor rather than the existing RSS-based AP selection.

Thus, in our opinion, it is important to generate algorithm that will not surpress the wireless network with congestion. So, it was concluded that the idea to generate the balancing AP selection mechanism is by having to consider all other performance metrices such as utilization, the loads and other relevant metrices rather than only considering the signal strength only. Besides that, the idea about the socialities and localization are also have to take into account in which these two factors are also major factor that contributed into congestion in the wireless LAN. On the other hand, the algorithm that any researchers made must consider the throughput of STAs present, load of traffics, socialities, localization and all other major factors of performances. Lastly, it is decided that, there must not be an intervention of NMS in the approach made as it can overload the network with traffics and congestion.