1 / 15

Potential Throughput Based Access Point Selection

Potential Throughput Based Access Point Selection. Suhua TANG Noriyuki TANIGUCHI, Oyunchimeg SHAGDAR, Morihiko TAMAI, Hiroyuki YOMO, Akio HASEGAWA, Tetsuro UEDA, Ryu MIURA, Sadao OBANA Nov.3, 2010, Auckland APCC 2010. Outline. Research background Proposed AP selection scheme

hiroko
Download Presentation

Potential Throughput Based Access Point Selection

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Potential Throughput Based Access Point Selection Suhua TANG Noriyuki TANIGUCHI, Oyunchimeg SHAGDAR, Morihiko TAMAI, Hiroyuki YOMO, Akio HASEGAWA, Tetsuro UEDA, Ryu MIURA, Sadao OBANA Nov.3, 2010, Auckland APCC 2010

  2. Outline • Research background • Proposed AP selection scheme • Simulation & Testbed evaluation results • Conclusion and future work

  3. WiFi AP Research Background Correspondent node Content server • Ubiquitous wireless access • Public networks (WiMax) • Private networks (Wireless LAN, Femtocell) • Users mutually share WiFi APs※  FON [1] • Project at ATR • Improve system throughput & reliability via cooperation among public/private networks • Cooperation among Wireless LAN/WiMax • multipath load balancing (WiMaxWiFi) • Operation in Wireless LAN (support more traffic) • APs power on, channel assignment • APs conduct power control to reduce co-channel interference • Nodes find APs via beacons • Nodes select AP (metric) • Nodes associate with selected AP (L2, association) • Nodes determine IP address (L3, MIPv6) • Channel access: CSMA/CA CN WiMax BS Internet AP1: … AP1: … AP2: … ※AP: WiFi access point

  4. AP Selection Problem • A node finds a path to talk with CN • The routing problem • AP-CN, broadband cables (fiber cable, VDSL etc) • Seldom be bottleneck • Node-AP, last hop wireless link • Poor quality of wireless link  poor performance • Assume the path between AP-CN is not a bottleneck  The routing problem is simplified to AP selection Internet CN Correspondent node Content server AP1 AP2 node i

  5. AP Selection Methods • Preparations: find APs • [2], Initial scanning & background scanning supported by driver • [3], SyncScan • Assume AP info available & focus on AP selection • Dealing with poor link quality • Selecting AP with RSSI※(WLAN driver) • →many nodes associate with a single AP • →load unbalance • →total throughput is degraded • Dealing with poor link quality & load unbalance • [4] (IEICE07), consider #associated nodes and link quality • [5] (MobiSys07), actually measure throughput • [6] (MSWiM08), estimate throughput via collision rate • Proposed scheme • Consider both channel congestion degree and link quality  potential throughput CN Correspondent node Content server Internet AP1 AP2 50% 70% 50% 30% 6Mbps 54Mbps node i ※RSSI:Received Signal Strength Indicator

  6. 20Mbps PT (Mbps) 10Mbps 6 12 24 54 Transmit rate (Mbps) E Concept of potential throughput ※ATR:air-time ratio • Air time ratio is used to indicate channel congestion degree. • A、B、C、D communicate with AP1 • ATR※=60% channel is used the percentage of busy time (60%) the percentage of idle time (40%) B A • Potential throughput for a new node E C • PT is for non-saturation scenario. • Node estimate PT→use idle channel→avoid overloading • When channel becomes saturated, throughput depends on contention. PT becomes similar to RSSI. D AP1

  7. Data (tData) ACK (tACK) L DIFS SIFS preamble header PSDU FCS preamble address FCS Definition of potential throughput • Potential throughput(PT): maximal throughput achieved by using all the idle channel and highest available rate → depending on channel availability & link quality • Channel congestion degree  available channel share AP STA rAP,STA • Link quality (per-bit transmission time): • tOH1: rate irrelevant overhead such as preamble • tOH2: rate dependent overhead such as MAC header • tOH1& tOH2include overhead of ACK • Lis payload length of typical packet (E.g. 512): unit: byte • PT between AP and STA 0<ATRm<1

  8. 2 1 1. measure ATR (ATRAP) 3. measure ATR (ATRSTA) 4. process beacon frame AP: ATRAP, RSSIAP 2. beacon (ATRAP) 5. calculate PTAP-STA 6. select an AP 7. association request Flowchart of AP selection • Each AP monitors channel & calculates ATRAP. • Each AP appends ATRAP in beacon. • STA scans channels & measures ATRSTA of the channel around itself. • On receiving beacon frames, STA detects the RSSI, infers the transmit rate, and finds ATRSTA from the beacon frame • STA calculates PT for each of the available APs. • STA chooses the AP which maximizes its PT. • STA associates with the selected AP. Ch1 Ch2 802.11a 802.11a AP STA (node) 50% 70% 50% 30% 6M 54M PT1=3M PT2=16.2M STA

  9. Parameter ATRm • Turn on STAiin order • Start the CBR※ flows between STAiand CN (via AP1) • CBR rate: the same for all flows • CBR rate: 1, 2, 3, 4, 5Mbps • Measure ATR at AP2 ※CBR: Constant bit rate ATR: reflect physical carrier sensing Not including DIFS/SIFS/BO

  10. Simulation Evaluation • Using network simulator Scenargie • STE (Space-Time Engineering) company’s product • http://www.spacetime-eng.com/ • Can be regarded as a successor of Qualnet in some degree • Include WiFi & WiMax • New functions implemented by STE too • Channel assignment, power control, AP selection, etc • Simulation setting • IEEE 802.11a parameter • Total throughput under different traffic unbalance degrees

  11. Simulation Results • 3 APs on orthogonal channels, connected to router via cables. • n1 STAs around AP1, n2 STAs around AP2, n3 STAs around AP3. • n1+n2+n3 is fixed at 12. • Nodes associate with an AP according to RSSI or PT. • One CBR flow per node, same setting for each flow • Channel is near saturation state when 4 flows share the channel. (n1, n2, n3)

  12. Simulation Results • Throughput gain depends on traffic disparity • Large gain when traffic is quite unbalance • Small gain when traffic is nearly balanced (n1, n2, n3)

  13. PT RSSI throughput ( 00:1B:8B:63:7C:78) AP1 RSSI: high PT: low STA (00:1B:8B:63:85:41) RSSI: low PT: high CN CN: correspondent node (00:1B:8B:63:85:7C) AP2 Testbed Experiments • Linux testbed with MadWiFi device driver (MadWiFi 0.9.4). • Add a new information element, ChInfo (carrying ATR value), to beacon frame. • A node records ATR in its scanning table on processing the ChInfo element • A node uses the calculated PT in the AP selection. • Experiment setting • Two APs, AP1 and AP2 run on different channels. • Channel around AP1 is more congested. AP1 has a higher ATR than AP2. • A new node, STA, nearer to AP1 and farther from AP2, is turned on. • Result: RSSI & PT; Throughput & PT.

  14. ( 00:1B:8B:63:7C:78) AP1 RSSI: high PT: low STA (00:1B:8B:63:85:41) RSSI: low PT: high CN CN: correspondent node (00:1B:8B:63:85:7C) AP2 Testbed Experiments • Iperf (UDP) is used to measure average throughput (30s period) between STA and CN. • Throughput gain depends on the disparity between channel congestion degree and link quality

  15. Conclusion • Our project focus on the cooperation of public (WiMax) & private (Wireless LAN) networks • This talk focused on AP selection • Channel congestion degree & link quality • Evaluation results • Improve per-node throughput & total throughput • Throughput gain depends on traffic unbalance degree • Future work • Evaluation of system throughput when all modules run together • Joint design of different modules

More Related