First of all, I would like to acknowledge the tremendous guidance, support and encouragement from my supervisor, Prof. Global Positioning System Graphical User Interface Independent Basic Service Set Industrial, Scientific and Medical Local Area Network.
LIST OF SYMBOLS
Chapter 1
INTRODUCTION
- Wireless Local Area Networks
- Channel Assignment in WLAN s
- Contributions of the Thesis
- Thesis Overview
Fitch, "Performance of Asynchronous Channel Assignment Scheme in Non-Uniform and Dynamic Topology WLANs," in Proc. The minimum neighbor with extended Kalman filter estimator (MINEK) channel assignment scheme is presented in Chapter 4.
LITERATURE REVIEW
IEEE 801 Standards
- WLAN Architectures
- MAC Protocol
- Performance Analysis of DCF
If the transmission is successful, a transition to the top of the Markov chain follows, and the process is repeated from the beginning for the next packet. Finally, at the maximum backoff stage, which corresponds to the maximum retransmission count, the packet will be discarded if the transmission still fails.
High Density WLAN s
It was found that in most cities, several hundred APs suffer from interference from at least three other neighboring APs. This is especially true because most APs have a significantly larger number of interfering neighbor APs.
Centralized Channel Assignment Schemes
- Utilization MinMax Scheme
- Degree of Saturation Scheme
- Degree of Saturation with Cost Scheme
Therefore, it is certainly interesting to evaluate the performance of the Uminmax scheme in these representative networks. It was found that the choice of the first assigned channel significantly affected the quality of the channel assignment.
Distributed Channel Assignment Schemes
- Weighted Coloring Scheme
- Communication-Free Learning Scheme
- Local Throughput Maximization Scheme
Consequently, the weights assigned will mostly reflect the interference experienced by customers who have only low traffic. The objective is to improve the total throughput of the network by maximizing the local throughput at each AP. For this reason, each AP maintains a table that keeps track of the number of clients of neighboring APs and the channel they are currently occupying.
The optimal value of the switching probability was found to be n = 0.5, which gave the best overall throughput in the shortest convergence time.
Summary
MINIMUM NEIGHBOUR SCHEME
Minimum Neighbour Scheme
- Minimum Neighbour Algorithm
- Equal Candidate Channels
- Asynchronous Operation
It is shown that to maximize throughput, each AP needs to select only the channel with the minimum number of neighbor nodes. This means that to maximize throughput, APs only need to select the channel with the minimum number of neighbor nodes. Assume that the number of channels is greater than the number of neighboring APs, that is D >.
During the first iteration, the AP-i runs the MINE scheme once and determines the channel with the minimum number of neighboring nodes.
Performance Evaluation
- Basic Performance
- Number of Channels
- Deployment Densities
- Non-Uniform Topologies
- Dynamic Topologies
- Heterogeneous Networks
Number of iterations and channel switches required for convergence for the MINE scheme and the LS scheme. 3.13, the total throughput of the MINE scheme and the LS scheme for A= 0 (static topology) and A= 2 (dynamic topology) for IEEE 802.1lb is shown. The throughput decomposition gains for the APs implementing the MINE scheme and the remaining APs (random scheme) are shown in Fig.
Similar to the case where only a percentage of APs implement the MINE scheme.
Summary
In addition, a throughput increase of at least 50% can be observed for the AP scheme of MINE at all PM values. Finally, the presence of APs implementing the MINE scheme results in increased throughput for APs implementing other channel allocation schemes such as the DSATUR scheme and the random scheme. This is because the MINE scheme increases its own throughput by reducing interference, which also benefits neighboring access points.
Therefore, due to its low complexity and strong performance, the MINE scheme is very attractive from a practical point of view.
MINIMUM NEIGHBOUR WITH
EKF ESTIMATOR SCHEME
Estimation of the Total Number of Nodes
The strategy for estimating the number of neighboring nodes starts with estimating the total number of nodes. Second, the knowledge of the estimated total number of nodes can be used in load balancing and handover algorithms for better network utilization [75-78]. First, it does not effectively capture the changes in the number of nodes when any AP switches in or out of the channel.
Given a certain level of conditional collision probability (which can be measured by observing the channel), function g(p) allows the determination of the corresponding number of nodes.
Estimation of the Number of Neighbour Nodes
After obtaining an estimate of the total number of nodes nk, the estimated number of nodes by AP, nk.i is derived. The implication of this fact is that the number of nodes of each AP is proportional to the average number of successful transmissions it makes. Consider nk nodes competing for a channel and let y; be the ratio of the number of successful transmissions by any AP to the total number of successful transmissions, y; = s/S.
Furthermore, whenever any AP switches into or out of the channel, the estimated net difference in the number of nodes can be defined as .
Minimum Neighbour with EKF Estimator Scheme
Using the new measurement, the EKF estimator produces an updated estimate of the total number of nodes, iik (Rules 1-2). From the perspective of API, it first obtains an estimate of the total number of nodes. Next, using this estimate, it distributes the total number of nodes to each AP.
Accurate estimates of the number of nodes from the AP for all APs are obtained from the API, as shown in Fig.
Performance Evaluation
- Upper Bound Performance
- Normalized Density
- Non-Saturated Load
- Unequal Load
- Fairness
- Scalability
The performance of the MINEK scheme compared to both upper bounds IS shown in Fig. It can be seen that the performance of the MINEK scheme is very close to the ideal interference-free case. The performance of the MINEK scheme is almost identical to that with the number of neighbor nodes being perfectly known.
These results show that the MINEK scheme can provide a significant improvement in throughput compared to other channel allocation schemes.
Summary
The EKF estimator and the number of nodes according to the AP estimation are proposed in this chapter, thus creating the MINEK scheme. Extensive packet-level simulation results show that the MINEK scheme can provide significant throughput improvements over other channel assignment schemes. Furthermore, it showed negligible performance difference when compared to the case where the number of neighboring nodes is known a priori.
Furthermore, the performance of the MINEK scheme is stable over a wide range of unsaturated and saturated load conditions.
CLIENT ASSISTED MINIMUM CONFLICT PAIRS SCHEME
Impact of Interference on Throughput with Multiple APs
5.l(a), when there are Trn (ciA, ap1) transmissions, AP2 and some clients in region 0 become hidden nodes. For Trn (cl0 , api) transmissions, some of its own clients in region A become hidden nodes. For Class II topologies, hidden nodes are reduced if there are fewer clients in region 0.
The total throughput of Class II topologies in relation to the total number of clients in region 0 is shown in Fig.
Minimum Conflict Pairs Scheme
Therefore, the number of conflict pairs for these topologies is the same as that for Class-I, as given in (5.2.1). Therefore, the number of conflict pairs for Class-II topologies with parallel transmissions can be given as. Therefore, the number of conflict pairs is simply the product of the clients of each BSS located in region 0.
For each of the interference classes, it can be seen that the throughput obtained is inversely proportional to the number of conflict pairs.
Performance Evaluation
- Number of Conflict Pairs
- Throughput
For both schemes, it can be observed that the number of conflict pairs increases as the client spread increases. The number of conflict pairs for both the MICP A scheme and the MINE scheme with different deployment densities for rEEE 802.11 a with high client proliferation is shown in Fig. First, it can be observed that the MICPA scheme outperforms the MINE and random schemes. all levels of client proliferation.
The throughput of the MICPA, the MYN and the random schemes with different deployment densities for IEEE 802.lla with high client distribution is shown in Fig.
Summary
This effectively reduces the performance gain of the client-backed MICPA scheme over the AP-only MINE scheme. Furthermore, significant gains of more than 10% and 40% in throughput were achieved over the AP-only MINE and the random schemes, respectively. On the other hand, the effects of client distribution and deployment densities on the performance of the MICP A scheme were also investigated.
It was found that the MICP A scheme is best deployed in scenarios with high customer dispersion and moderate deployment density (normalized density close to unity).
CONCLUSIONS AND FUTURE WORK
Conclusions
It was shown that to maximize throughput, each AP only needed to select the channel with the minimum number of active neighboring nodes. The implementation of the MINE scheme in practice requires the estimation of the number of neighboring nodes. In the MINEK scheme, an EKF estimator was formulated to estimate the total number of nodes.
Extensive simulation results have shown that the MICPA scheme was able to provide a significant reduction of up to 80% in the number of conflict pmrs.
Suggestions for Future Work
Therefore, it is expected that multiple channel allocation schemes will have to coexist within the same network. To the best of our knowledge, the interaction between different channel assignment schemes has not yet been investigated. These studies can ensure a truly harmonious coexistence between different channel allocation schemes for the betterment of all parties.
While distributed channel assignment schemes can be implemented in either case, larger deployments have opportunities to leverage the communication and collaboration between the APs they own.
Barbosa, “Performance evaluation of automatic channel assignment mechanism for IEEE 802.11 based on graph colors,” in Proc. Tinnirello, “Kalman Filter estimation of the number of competing terminals in an IEEE 802.11 network,” in Proc. Vitsas, “Optimization of RTS/CTS handshake in IEEE 802.11 wireless LANs for maximum performance,” in Proc.
Wong, "Back-of-the-Envelope Computation of Throughput Distributions in CSMA Wireless Networks," in Proc.
APPENDIX A
OPNET SIMULATION DEVELOPMENT
Before any modification could be attempted, a clear understanding of the node model and associated process models was necessary. Implementing, debugging and verifying the required new features and the MINEK scheme in the model took considerable effort and time. Calculated number of empty slots based on elapsed time divided by slot time.
Obtained BSSID from packet header. i) Started the EKF estimator when a new Pk measurement was available. ii) Checked if channel change occurred and determined uk, estimated net difference in number of nodes. iii) Check if an alarm has occurred and set the Qk variance accordingly. iv) All EKF equations are implemented to provide an updated estimate of nk. v) All change detection filter equations are implemented, setting an alarm if necessary. v) They added up the number of successful transfers of all APs and determined whether they exceeded the given threshold. vi) Implemented equations that provide an updated estimate of nk.i. vii) A specified number of neighbor nodes for all available channels and the channel with the least number of neighbor nodes. viii) Checked that all conditions for channel change are met. ix) Changed channel and reset relevant parameters.
