1. Network Research Center Tsinghua Univ. Beijing, P.R.China
Performance Optimization in Wireless Mesh Networks Based on Mesh Point Priority Mechanism
Rongdi Chen
Network Research Center
Tsinghua Univ. Beijing, P.R.China

2. Outline Background Performance degradation in WMN
Mesh Point Priority Mechanism
distinguish transmit priority of mesh points from two perspectives
Performance evaluation
Future works
Conclusions
Background
WMN obtains great consideration for its advantages and widely usage scenarios
Current WMN solutions use standard MAC protocol and it is not suitable for backhaul networking, especially in single radio single channel situation
Performance degradation in WMN (2 figures)
WLAN: deteriorate by the number of competing stations increases
WMN: deteriorate when the number of WMN hops or the number of competing stations increases
How to solve these problems?
Number of WMN hops: increase number of radios to make backhaul networking can send and receive simultaneously
Number of competing stations: adjust the parameters of MAC Protocol
Former work: Bianchi, Hui Ma change the sending probability
Use it in WMN
Mesh Point Priority Mechanism: distinguish transmit priority of mesh points from two perspectives
Performance evaluation
Simulation setup
Simulation results and analysis
2019/4/21

3. Facilitate deployment
Background
Internet
Internet
Gateway
Internet
Gateway
Extend coverage
Facilitate deployment
2019/4/21

4. Background Cont.
2019/4/21

5. Current Status of WMN Several mesh network solutions
Nortel WMN
Tropos MetroMesh
Azalea MSR
MeshDynamics Structured Mesh ……
More real application scenarios
Wireless campus
Wireless city: Philadelphia, New York, Kaohsiung, Taibei ……
Solutions state-of-the-art? No!
The use of standard MAC protocol
Not suitable for the backhaul networking in WMN: media sharing and the weakness of CSMA/CA
Performance deteriorates significantly: when the number of WMNs hops or the number of competing stations increases, especially for single radio single channel situation
2019/4/21

6. Performance Degradation in Wireless Mesh Network
Performance degradation: single radio single channel
Effective system throughput: the total quantity of UDP
payload received by all the destinations within a second
2019/4/21

7. How to solve these problems?
Degradation by number of hops: increase radios
Use two or more radios dedicated for backhaul
Degradation by number of competing stations: adjust protocol parameters
Bianchi: CWmin has great influence on the system performance
Performance evaluation and enhancement of the CSMA/CA MAC protocol for wireless LANs
IEEE saturation throughput analysis
Hui Ma: Optimize multiple competing stations throughput degradation in WLAN (1 hop) situation by adjusting CWmin values
Range estimation and performance optimization for IEEE based on filter
Does it work well in multihop situation?
Question to be answered in our paper
2019/4/21

8. Analysis
WMN: backhaul networking, packet transmit in hop by hop fashion
Mesh Access Point VS Normal station
Same access opportunity to wireless medium
Different contribution to effective system throughput
the data transmitted by MAPs contribute more to effective system throughput than by normal stations, as the data should be transmitted by MAPs before they arrive at the destination
Among different hop MAPs, the contribution to effective system throughout is also different
We should distinguish these differences and guarantee the priority
How to?: give MAP higher sending probability by adjusting CWmin and CWmax
2019/4/21

9. Mesh Point Priority Mechanism (MPPM)
MPPM distinguishes transmit priority of mesh points from two perspectives:
The MAPs should have higher priority than the normal stations due to their outstanding contribution to effective system throughput.
The priority among the MAPs should be adjusted according to the topology and traffic load of the wireless mesh network.
Change the priority by adjusting the MAC protocol parameters: CWmin & CWmax
Smaller contention window represents higher sending probability
We can set smaller CWmin and CWmax for MAPs to get higher priority
2019/4/21

10. Performance Evaluation
NS2 simulator with the CMU wireless extensions and NOAH routing protocol
Simulation parameters accord with IEEE b network specification
Ideal channels
RXThresh_: 5.844e-9, 125m
CSThresh_: 3.652e-10, 250m
dataRate: 11Mbps
UDP packet size: 1500 bytes
Simulation time: 400s
Saturated data source
CWmin_
Normal Stations
MAPs 31 15
CWmax_
1023
2019/4/21

11. Simulation results (1) 2 Hop Throughput 2019/4/21
All the stations associated to 2 Hop MAP
Simulation runs for 400s
Calculate effective system throughput within each second
Use the steady data set to calculate Trimmed Mean throughput
Illustrated with MATLAB
By using MPPM the effective system throughput is greatly improved especially when the number of competing stations increases
2833.3Kbps
12.9%↑
Kbps
2510.5Kbps
138.4%↑
Kbps
2 Hop Throughput
2019/4/21

12. Simulation results (2) 3 Hop Throughput 2019/4/21
All the stations associated to 3 Hop MAP
Simulation runs for 400s
Calculate effective system throughput within each second
Use the steady data set to calculate Trimmed Mean throughput
Illustrated with MATLAB
By using MPPM the effective system throughput is greatly improved especially when the number of competing stations increases
1552.3Kbps
1150.0Kbps
80%↑
862.4Kbps
180.3%↑
410.3Kbps
3 Hop Throughput
2019/4/21

13. Simulation results (3) 3 Hop Throughput Case 1 2019/4/21
One station is associated to 2 Hop MAP and sends UDP flow to Server S at the rate 6 Mbps
The other stations are associated to 3 Hop MAP
Simulation runs for 400s
Calculate effective system throughput within each second
Use the steady data set to calculate Trimmed Mean throughput
Illustrated with MATLAB
By using MPPM the effective system throughput is greatly improved especially when the number of competing stations increases
1855.6Kbps
1413.0Kbps
92.0%↑
966.8Kbps
140.8%↑
586.7Kbps
3 Hop Throughput Case 1
2019/4/21

14. Simulation results
The situation that many stations are associated to both 2 Hop MAP and 3 Hop MAP is also considered. The simulation result shows that MPPM still functions well.
So in WMN, the MPPM can gain much higher throughput than standard IEEE MAC
If we dynamically adjust the CWmin and CWmax for both MAPs and stations based on the topology and traffic load of WMN, the more throughput gains would be achieved.
2019/4/21

15. Future works
More precise way to map the priority to the parameter value
Ideal adaptive optimization mechanism
Consider the fairness issue in WMN
Achieve relative fairness by adjusting the protocol parameters
2019/4/21

16. Conclusions
Wireless Mesh Networks: a promising broadband wireless access infrastructure but suffers performance degradation problem
Degradation when the number of hops increases
Degradation when the number of competing stations increases
Mesh Point Priority Mechanism
Distinguish transmit priority of mesh points from two perspectives by adjusting the parameters of IEEE MAC protocol
Optimize effective system throughput
Simple but effective
Some future works
Ideal adaptive optimization mechanism
Consider fairness issues in Wireless Mesh Metworks
2019/4/21

17. Thank you!
Any idea or other
suggestions?
2019/4/21