Sie sind auf Seite 1von 5

Channel Assignment Strategies of IEEE 802.

11 WLAN in a Shared Public Hotspot Environment


Songrit Srilasak, Kitti Wongthavarawat, Kitiwat Limmongkol, Siwaruk Siwamogsatham
Wireless Innovation and Security Laboratory National Electronics and Computer Technology Center Pathumthani, Thailand be aware of the interference problem and may not give the full cooperation. Channel assignment of APs is the challenging problem in the shared public hotspot environment, especially in the situation that other service providers do not provide full cooperation. Channel assignment of APs need to be carefully determined to obtain the optimal performance. Most of previous study discusses the interference effect in assignment channel using mathematical model as in [2]. [3] discusses the interference among non-overlapping channel through the experimentation. This paper proposes the channel assignment strategies of IEEE 802.11 WLAN to provide the optimum service performance among shared service providers. The series of experiment presented in this paper provides the basis of the proposed the channel assignment strategies. Next section provides the overview of IEEE 802.11 specification in the part of channelisation and carrier sensing. Section III demonstrates the testbed and experimental result. Section IV provides the proposed channel assignment strategies. The summary is in Section V. II. IEEE 802.11 STANDARD

Abstract The demand of wireless access in anywhere and anytime drives the deployment of Wi-Fi hotspot. Multiple service providers may deploy the Wi-Fi hotspot in the same area to response to the users needs. Channel assignment for the access point becomes the challenging problem, especially in the situation that other service providers do not provide full cooperation. Channel assignment of APs need to be carefully determined to obtain the optimal performance. This paper proposes the channel assignment strategies of IEEE 802.11 WLAN to provide the optimum service performance among shared service providers. The series of experiment presented in this paper provides the basis of the proposed the channel assignment strategies. Keywords; Wi-Fi, IEEE 802.11 WLAN, Channel Assignment

I.

INTRODUCTION

IEEE 802.11 WLAN [1] or Wi-Fi is an attractive networking solution due to the ability of computing and communication devices (e.g., PDA, PC, notebooks) equipped with appropriate WLAN Network Interface Cards (NICs) to communicate with each other without encumbering the users with cabling and its limitations. Wi-Fi changes the way people work. It allows people to stay connected or share data anywhere, anytime. In public hot spots such as airports, bars and coffee shops, business owners are providing travelers and patrons with access to the Internet through public Wi-Fi. This gives business travelers high-speed access to their corporate networks and leisure travelers the chance to check their email and read the news. Such connections can also be established while a traveler sits in the airport lounge without a need for a cable based connection. At present increasing in the demand of public Wi-Fi access drives the wide deployment of Wi-Fi public hotspot. To provide the Wi-Fi public hotspot service, an access point (AP) needs to be installed to cover the area where most people expected to use. Some area is the common area (such as airport, shopping mall and etc) where multiple service providers provide the public hotspot services in the same area. Due to the shared medium of the Wi-Fi, the Wi-Fi signal from different service providers may provide the harmful interference to each other and then degrade the service quality. In order to reduce the interference and improve the service quality, the cooperation among service providers in channel planning is necessary. However, cooperation between service providers is not an easy task. Some service providers may not

The Institute of Electrical and Electronics Engineers (IEEE) has been developing IEEE 802.11 standards which contain specifications for several technologies, covering various speeds and frequencies. In June 1997, the first standard of IEEE 802.11 was approved. The standard details the Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications that operate over the 2.4 GHz Industrial, Scientific and Medical (ISM) frequency band supporting data rates of 1 Mbps and 2 Mbps. In 1999, IEEE approved two supplement standards: IEEE 802.11a and IEEE 802.11b. IEEE 802.11b enhances the speed to 5.5 Mbps and 11 Mbps using Complementary Code Keying (CCK) modulation in the 2.4 GHz band. These standards allow for multirate support and backwards compatibility. IEEE 802.11b supports speeds of 1, 2, 5.5 and 11 Mbps. IEEE 802.11a standard has an enhanced speed of up to 54 Mbps using Orthogonal Frequency Division Multiplexing (OFDM) radio technology in the 5 GHz band. IEEE 802.11a supports speeds of 6, 9, 12, 18, 24, 36, 48, and 54 Mbps. On June 2003 the IEEE 802.11g standard was approved. It uses OFDM based physical layer to achieve speeds of up to 54 Mbps in the 2.4 GHz band (same frequency

band as IEEE 802.11b). IEEE 802.11g provides speeds of 1, 2, 5.5, 6, 9, 11, 12, 18, 24, 36, 48 and 54 Mbps and is backwards compatible with IEEE 802.11b devices which operate in the same frequency range at speeds of 1, 2, 5.5 and 11 Mbps. A. Channelisation IEEE 802.11, 802.11b, and 802.11g support 11 channels (in North America) or 13 channels (in Europe) as shown in Figure 1. Each channel has 22 MHz Bandwidth and the channel spacing between two adjacent channels is 5 MHz. Therefore two adjacent channels are overlapping channels. Channel 1 and Channel 6 (separated apart by 25 MHz) are considered to be non-overlapping channel and so as Channel 6 and Channel 11. The real spectral mask of a channel defined in the standard is shown in Figure 2 (IEEE 802.11b) and Figure 3 (IEEE 802.11g). From the spectral mask, although channel 1 and channel 6 are 25 MHz apart and assumed to be nonoverlapping channel, the spurious spectrum of both channels are still overlapping. These affect the performance of both channels in some specific condition. Such the condition will be discussed in the next section. IEEE 802.11 WLAN devices can operate in the shared medium environment over the unlicensed ISM band. The standard defined the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) as the media access control mechanism to self-coordinate among IEEE 802.11 WLAN devices. Before a WLAN device transmits a packet, the carrier sensing mechanism is used to determine whether there are ongoing transmissions on the channel. If the channel is busy, the WLAN device will defer the transmission. On the other hand, if the channel is idle (no other ongoing transmission), the device will transmit the packet. However, it is possible that two WLAN devices are hidden to each other and both sense the channel is idle, and then transmit packets at the same time. The packets collision occurs. IEEE 802.11 standard defined the backoff mechanism to resolve the collision.

Figure 3. IEEE 802.11a/g Spectral Mask [1]

B. Carrier Sensing IEEE 802.11 standard defined two types of carrier sensing mechanisms, one is the physical carrier sensing and the other is the virtual carrier sensing. Both mechanisms determine the state of the channel (busy or idle). If either mechanism indicates the busy channel, the channel is considered to be busy. Physical carrier sensing is defined in Physical Layer Convergence Protocol (PLCP) of IEEE 802.11 PHY. Physical carrier sensing determines whether the channel is busy by 1) detecting the preamble of the IEEE 802.11 packet (Figure 4) and 2) evaluating the energy level of signal. Physical carrier sensing can be configured with several Clear Channel Assessment (CCA) options: CCA Mode 1: Energy above threshold. CCA reports a busy medium upon detecting any energy above the threshold. CCA Mode 4: Carrier sense with timer. CCA starts a timer whose duration is defined in the standard and reports a busy medium only upon the detection of a valid preamble of a packet until the timer expires. Otherwise it reports the media is idle. CCA Mode 5: A combination of carrier sense and energy above threshold.

Figure 1. IEEE 802.11 Channelisation

Virtual Carrier Sensing is defined in IEEE 802.11 MAC. Virtual Carrier Sensing carries out through Network Allocation Vector (NAV) indicating the time duration of the future transmission of a packet. NAV value is included in RTS/CTS packet and also all data packets. The duration field in IEEE 802.11 MAC header of a data packet determines NAV. NAV in the header of a transmitted packet serves as the notification to all other stations about the time duration of its packet transmission. Each station maintains its own NAV counter. Every time a station detects NAV of a packet in the channel, it will update NAV counter with the latest NAV. A station keeps counting down NAV counter until reaching zero, then the virtual carrier sensing of the station will indicate the channel is idle; otherwise, the channel is busy.

Figure 2. IEEE 802.11b Spectral Mask [1]

client MAC address. The unicast traffic data is sent to Client through TestAP. c) An access point as TestAP using off-the-shelf AP (Linksys WRT54GL). We flashed the TestAP with the Tomato firmware [6]. Tomato firmware provides all basic WLAN functionalities as in original Linksys firmware and additional Received Signal Strength Indication (RSSI) measurement capability. We use this capability to measure RSSI of ForeignAP experienced by TestAP. TestAP is set to IEEE 802.11g mode only. 3) Client notebook: Client associates to TestAP. DU Meter software [7] is installed to measure the throughput performance of the incoming UDP traffic over the wireless. 4) Anritsu BTS Master MT8222A [8] is a spectrum analyzer. The distance (d1) between TestAP and ForeignAP is varied depending upon the experiments while the distance (d2) between TestAP and Client is fixed at 2 meters. A. Experiment 1: Channel measurement. The experiment is aimed to measure the characteristic of IEEE 802.11 channelization as well as the background noise. Three ForeignAPs broadcast UDP traffic on different channels (i.e., channel 1,6,11). Spectrum analyzer measures the signal in two different locations: one location is very close to the ForeignAP, and the other is 20 meters away from ForeignAP. Figure 6 demonstrates the measurement result of channel 1, 6, 11 spectrum in the near and far measurement location. The background noise is measured when there is no AP operating. In the measurement far from ForeignAP, the spurious emission of all three channels is negligible. On the other hand, in the close measurement, the spurious emission of all channels is noticeable (the total of 4 dBm above the background noise). This spurious emission may affect the throughput performance of other APs.

Figure 4. PLCP Frame Format

III.

TESTBED AND EXPERIMENTAL RESULT

The result of the series of the experimentation in this paper provides the basis of the proposed channel assignment strategies. The testbed setup is shown in Figure 5.

Figure 5. Testbed Setup

The testbed is consists of 1) Three access points (AP) assigned as ForeignAPs using off-the-shelf AP (Linksys WRT54GL [4]). ForeignAP represents the access point from other ISP providing the public hotspot service in the same area. We configure ForeignAPs with the default value and assign the de-facto standard nonoverlapping channels (i.e., channel 1, 6, 11). 2) Two notebooks as Traffic Generator: packetETH [5] , an open source tool to create and send any packet to the Ethernet, is installed on the notebooks. a) Background traffic generator: packetETH generates UDP packet with 1470 byte packet size and with maximum transmission rate. The destination MAC address is set to unknown MAC address starting with (FF:X:X:X:X:X). By setting this, the UDP packet will broadcast through all three ForeignAPs with maximum transmission rate possible. This will provide the simulated scenario in which the public hotspot is very busy. b) Test traffic generator: packetETH generates UDP packet with 1470 byte packet size and with maximum transmission rate. The destination MAC address is set to the

Figure 6. Spectrum Measurement

B. Experimental 2: Maximum Throughput This experiment is aimed to measure the maximum throughput of TestAP and Client pair as a benchmark. There is no other AP operating. The testbed is located in the area with the minimum background noise (no other AP around). Test traffic generator sends UDP packets to Client through TestAP with 1470 byte packet size and with maximum transmission rate. The throughput is measured from DUMeter in Client. The maximum throughput of the system is around 29 Mbps in all channels. C. Experiment 3: Throughput performance in co-channel, overlapping channel and non-overlapping channel environment This experiment is aimed to investigate the throughput performance in co-channel, overlapping channel and nonoverlapping channel environment. Only one ForeignAP operates in the channel 1 with the background traffic. Test traffic generator sends UDP packets to Client through TestAP with various channels (i.e., channel 1-11). The throughput is measured from DUMeter in Client in different channels. Figure 7 shows the throughput performance comparing to the maximum throughput. The result illustrates the throughput is degraded on the co-channel (i.e., channel 1) and overlapping channel (i.e., channel 2,3). The throughput improves in channel 4 and reaches the maximum from the channel 5 and higher. Figure 8 demonstrates the spectrum measurement of channel 1,3 pair and channel 1,6 pair. The TestAP in channel 3 performs worst in the test due to the severe interference with the ForeignAP. Channel 1 (co-channel) performance is better than channel 2 and 3. The explanation is that in the co-channel environment all APs are able to receive each other packets as well as NAV in the packets. The co-channel APs can perform carrier sensing mechanisms well in both physical carrier sensing and virtual carrier sensing, while in the overlapping channel APs case, only physical carrier sensing is effective. Furthermore, with overlapping channel, the TestAP may not efficiently detect preamble of the packets of ForeignAP in other channels. CCA Mode 1 (Energy Above Threshold Mode) is only choice for the physical carrier sensing. Physical carrier sensing become less effective in this case. Therefore TestAP is prone to interference with ForeignAP. The support evidence is presented in the Figure 7. Let assume ch is the channel difference between two APs. The conclusion is that ch 3 provides the acceptable throughput performance. D. Experiment 4: Throughput performance in co-channel, overlapping channel and non-overlapping channel environment (ForeignAP and TestAP are in very close proximity). This experiment repeats the experiment 3. The only difference is ForeignAP and TestAP are in very close proxmity. RSSI of ForeignAP measured by TestAP is -14 dBm. The result is presented in the Figure 10. We performed this experiment in several iterations. The result is unstable. Even throughput of TestAP (in channel 11) also degrades. From this experiment, we conclude that the close installation of

APs causes unpredictable throughput performance of APs and it is not recommended.

Figure 7. TestAP Throughput co-exists with ForeignAP Channel 1

ForeignAP-CH1

TestAP-CH3

Figure 8. Channel Spectrum of ForeignAP-CH1 and TestAP-CH3

ForeignAP-CH1

TestAP-CH6

Figure 9. Channel Spectrum of ForeignAP-CH1 and TestAP-CH6

C. Algorithm If there is Ch(a) in which |Ch(a) Ch(f)| 3 for all Ch(f), then, assign Ch(a) such that else, assign Ch(a) = Ch(f) such that foreign AP channel has lowest activities.
Figure 10. TestAP Throughput co-exists with ForeignAP Channel 1 (Very close proximity)

Max

all Ch(f)

Ch(a) Ch(f)

E. Experiment 5: Throughput performance in the simulated public hotspot environment. This experiment is aimed to get insight of how to set the channel in the public hotspot environment with the best performance. ForeignAPs operates on channel 1, 6, 11 with background traffic. The distance between ForeignAP and TestAP is around 20 meters. RSSI of ForeignAP measured by TestAP is around -70 dBm. We perform the throughput measurement of TestAP from various channels (1-11). The result is in the Figure 11. In this case, there is no nonoverlapping channel available. The throughput performance of co-channel (i.e., Ch1-Ch1 pair, Ch6-Ch6 pair or Ch11-Ch11 pair) is better than all other overlapping channels. The result is counterintuitive so that co-channel assignment is better than overlapping channel assignment. IV. PROPOSED CHANNEL ASSIGNMENT STRATEGIES

Figure 11. TestAP Throughput in simulated shared hotspot environment

V.

SUMMARY

The experimentation in previous section provides the basis of the proposed channel assignment strategies. The proposed channel assignment strategies provide the guideline of how to set our AP in the public hotspot environment to obtain the optimum performance. A. Assumptions 1) In the public hotspot environment, we could not control the channel setting or the location of other APs (i.e., ForeignAP). 2) The transmitted power of all APs follows the FCC rule part 15. 3) Our assigned AP locates away from other APs at least 10 meters. 4) Public hotspot channel is busy. B. Definitions 1) Ch(a) = assigned channel 2) Ch(f) = channel of foreign AP in which the signal strength is greater than -90 dBm at our assigned AP location. 3) Ch(a), Ch(f) {1, 2, .., 11} 4) |Ch(a) Ch(f)| = absolute value of channel difference between assigned channel and foreign AP channel

The demand of wireless access in anywhere and anytime drives the deployment of Wi-Fi hotspot. Multiple service providers may deploy the Wi-Fi hotspot in the same area to response to the users needs. Channel assignment for the access point becomes the challenging problem, especially in the situation that other service providers do not provide full cooperation. Channel assignment of APs need to be carefully determined to obtain the optimal performance. This paper proposes the channel assignment strategies of IEEE 802.11 WLAN to provide the optimum service performance among shared service providers. The series of experiment presented in this paper provides the basis of the proposed the channel assignment strategies. References
[1] IEEE 802.11 WG, IEEE Std 802.11-2007: IEEE Standard for Local and Metropolitan Area Network Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications R. Akl, A. Arepally, Dynamic Channel Assignment in IEEE 802.11, PORTABLE07 P. Fuxjager, D.Valerrio, F.Ricciato, Myth of Non-Overlapping Channel: Interference Measurement in IEEE 802.11, WONS apos07

[2] [3] [4] [5] [6] [7] [8]

http://www.linksys.com http://packeth.sourceforge.net/ http://www.polarcloud.com/tomato http://www.dumeter.com/ http://www.us.anritsu.com/BTSMaster

Das könnte Ihnen auch gefallen