Download presentation
Presentation is loading. Please wait.
Published byMartin Bryan Modified over 9 years ago
1
Dynamic Load Balancing through Association Control of Mobile Users in WiFi Network Presenter: Chia-Ming Lu Huazhi Gong, Student Member, IEEE, Jong Won Kim, Senior Member, IEEE IEEE Transactions on Consumer Electronics, 2008
2
Abstract Association between mobile users (MUs) and access point (AP) is based on the signal strength information Extremely unfair bandwidth allocation among MUs Propose a distributed association algorithm Achieve load balancing among the APs 2
3
Outline Introduction IEEE 802.11 Basics System Model and Problem Definition Distributed Association Algorithm Performance Evaluation Conclusion 3
4
Introduction IEEE 802.11 MAC has an “performance anomaly” Data rate information is required to guide load balancing schemes 4
5
Introduction(cont.) Default best-RSSI(receiving signal strength indicator)- based AP selection scheme Result in severe unfairness and even poor overall performance 5
6
Introduction(cont.) AP selection algorithms can be divided into two categories: Centralized optimization NP-hard nature for performing the centralized computations Distributed heuristic methods Do not consider the multiple data rate information or propose non-practical solutions The proposed scheme Evaluated the performance by the numerical simulator Realistic scenario with mobility pattern Implemented a prototype on small-scale testbed 6
7
IEEE 802.11 Basics Distributed Coordination Function (DCF) Association procedure for the roaming mobile user(MU) 7
8
System Model and Problem Definition 8 θ: MU throughtput U: MUs L: packet length y a : AP load d: time required to transmit one packet from MU r: physical data rate of MU m: number of retrials required to transmit one packet
9
Distributed Association Algorithm Association Algorithm for APs and MUs 9
10
Performance Evaluation Jain’s fairness index Numerical Simulation for Realistic Scenario – large scale Packet Level Simulation – medium size Prototype Implementation – small scale : total throughput 10
11
Fig. 4. The snapshot of developed numerical simulator. SimPy Performance Evaluation(cont.) Numerical Simulation for Realistic Scenario 11 Fig. 5. A realistic scenario with measured mobility for numerical simulation. The red squares denote the APs and the blue circles denote the MUs at the beginning of simulation. 56 APs and 126 MUs 1100x1000m 2
12
Performance Evaluation(cont.) Fig. 6. The throughput difference between RSSI-based scheme and proposed scheme 12
13
Performance Evaluation(cont.) Fig. 7. The Jain’s fairness value difference between RSSI-based scheme and proposed scheme 13
14
Performance Evaluation(cont.) Packet Level Simulation NS2 simulator 9 AP and 40 MUs 600×600m 2 TABLE I COMPARISON FROM NS2 SIMULATIONS 14
15
Performance Evaluation(cont.) Prototype Implementation MadWifi-ng wireless driver Two Dell laptops as Mus Two Dell Desktops as APs 15
16
Performance Evaluation(cont.) 16 Fig. 10. The measurement results to compare the performance difference between the default RSSI-based scheme and the proposed scheme
17
Conclusion The load balancing scheme Guarantee the throughput fairness among the MUs Gradually balances the AP loads Feasibility of the proposed scheme Modifying open source wireless driver Achieve apparent load balancing 17
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.