1. IEEE 802.11 Wireless LAN Part II Access Point, Power Management, Polling, and Frame Format 14-1

2. Access Points Used in the infrastructure mode. Functions like a bridge between an IEEE 802.11 wireless network and an Ethernet. An infrastructure mode station must associate with an access point. –All frame transfer must pass through the AP even if frames can be directly sent to other stations using the ad hoc mode. The AP, acting like a central controller, thus can provide some management (e.g., Power management) and QoS (e.g., PCF) services that cannot be achieved by DCF. 14- 2

3. Beacon Generation The AP periodically generate and broadcast beacon frames. (In an ad hoc network where no AP is present, stations take turns to generate beacons.) The beacons act as a one-way communication channel. They convey and announce important information to all associated stations. For example: –By listening to the air and searching beacons, a station can know whether there is an AP nearby. –The beacons can announce the AP’s identity, capability, and current timestamp. –They can be used to synchronize the clocks of all stations. They can also be used to support contention free frame transfers. (provide better QoS) 14- 3

4. Beacon Format 14- 4

5. Beacon Generation in an Infrastructure Network Beacons should be generated and transmitted at the “beacon interval.” Before they are sent, they should follow the basic access method defined in DCF. I.e., they need to wait for a DIFS idle period. They may be delayed if the medium is busy at that time. 14- 5

6. Beacon Generation in an Infrastructure Network The AP and each station maintains a clock. The AP acts as the timer master, all stations that receive the beacons generated by the AP should set their clocks to the AP’s clock. Why synchronizing stations’ clocks? –Frequency hopping needs time synchronization. –Periodic wakeup in the power save mode also requirea that all doze-mode stations wake up at the same time. 14- 6

7. Beacon Generation in an Ad Hoc Network Stations use a distributed algorithm to take turns to generate beacons. TBTT TBTT: target beacon transmission time. 14- 7

8. At TBTT, the backoff timer for any pending non- beacon frame is suspended. Calculate a random delay between [0, 2*Cwmin] Wait for the period of the random delay to expire. If a beacon arrives before the random delay timer expires, the remaining timer is canceled. We do not need to transmit a beacon. Otherwise, if the random delay timer has expired and no beacon has arrived during this period, we send a beacon. The Distributed Algorithm Used in the Beacon Generation in an Ad Hoc Network 14- 8

9. Scanning AP Procedure When a station is powered up, it needs to find out in which frequency channel there is a desired AP. (If there is one, the station wants to get associated with it.) The process of finding AP is called scanning. There are two different types of scanning: –Passive scanning Listen to each channel for some time to see whether we can hear some beacons. –Active scanning Sending probe request frames actively on each channel. AP, on receiving a probe request frame, must return a probe response frame. The station thus can more quickly know where there is an AP. 14- 9

10. Active Scanning Procedure On hearing a broadcast probe request frame, before sending back their response frames, APs should use the DCF method to avoid collisions. 14- 10

11. Start or Join a BSS With the information gathered from the passive or active scanning, a station can decide to: –Start a BSS by itself (if there are no existing BSSs) It can form an ad hoc network. –Then the ID of the new formed BSS is a locally administered MAC address that is randomly chosen. It can act as an AP and form an infrastructure network. –Then the ID of the new BSS should be the MAC address of the AP. –Join an already existing BSS If the BSS is an infrastructure network, the station needs to send association request to the AP. 14- 11

12. Point Coordination Function PCF provides contention free frame transfer. Through its polling mechanism, QoS may be achieved. PCF works only in an infrastructure network and the PC reside in the access point. PCF frames have a high priority than normal DCF frames. Thus PCF can control the use of the medium. –PCF frames need to wait PIFS time before they are sent. –PIFS is shorter than DIFS, which is used by normal data frame transfer. PCF is an option. So far, no product in the market supports it. 14- 12

13. Point Coordination Function The PCF divides the time into a series of CFP intervals. Each CFP interval is further divided into a contention free period (CFP) and contention period (CP). Inside the CFP, the PC polls stations to give them a guaranteed chance to send out their frames. Need to wait only PIFS. Therefore, the PC can reserve the medium. All stations should set their NAV so that they are not allowed to use DCF to use the medium during this period. 14- 13

14. CFP Starts and Repeats at DTIM A DTIM (delivery traffic indication message) interval in this case is composed of 4 beacons. A CFP repetition interval (contention-free period) in this case is composed of 3 DTIM. CFP_Dur_Remaining is updated in each successive beacon. BB NAV reserves the medium 14- 14

15. Shortened CF MaxDuration Because beacon generation and transmission may be delayed by unfinished DCF traffic (they use SIFS.), the contention free period may need to be shortened. This change will be announced in beacon frames. 14- 15

16. Polling In a CPF, the PC sends poll frames to stations asking whether they have frames to send. If a polled station has a frame to send to the AP, it can send it to the AP now. 14- 16

17. Polling To reduce polling frame overhead, the AP should piggyback the data, poll, and ACK as mush as it can onto a frame. Inside the CFP, frames are exchanged with a SIFS gap. This allows them to finish without being interrupted by traffic from other stations. ACK 14- 17

18. Polling If there is no response from a polled station, the PC starts polling again after PIFS. The PC can end the CPF earlier by sending out CF- END if there is no more station to poll. All other stations should also update their NAV. 14- 18

19. Polling Is Not Always Needed Actually in a CFP, the PC does not need to always poll stations. Polling stations give stations a chance to send (uplink) their frames to the AP. However, if there is no uplink traffic and the AP has many frames to send to stations (downlink), it can send these frames to stations without polling them. Therefore, to just receive frames from the PC, stations need not be CF-pollable. 14- 19

20. Polling Policy Polling list processing: –The PC should send a poll to at least one station during each CFP when there are entries in the polling list. –The PC should issue polls to stations based on their AID (association ID) in ascending order. –In a CFP, if all stations in the list have been polled and time still remains, the PC can issue polls to any station on the list. –Or, the PC can send data or management frames to any station. Still a research issue 14- 20

21. Power Management Wireless NIC usually are used by mobile devices. (for mobility) Mobile devices usually uses battery power, which cannot last too long. Wireless NIC can consume a lot of power while waiting for frames to arrive. (CISCO Aironet: Transmit: 450 mA, Receive: 270 mA, Sleep mode: 15 mA ) Therefore, to save power, IEEE 802.11 defines a power save mechanism by which wireless stations can sleep most of the time and be waked up when it has a frame to receive. 14- 21

22. Power Saving Scheme The general idea is as follows: –A station indicates in its transmitted frame to the AP that it will sleep after this frame transmission. –Then the station goes to sleep. –In the mean time, if there is a frame that the AP needs to forward to the station, the frame will be buffered at the AP. –In the beacons generated by the AP, there is information showing that there is a frame buffered for the station. –Because the sleeping station needs to wake up periodically to receive the AP’s beacons, it will notice this. –It then sends a power-save poll request frame to the AP. –Then the AP will send the buffered frame to the station. –Then the station can sleep again. 14- 22

23. Power Save in Infrastructure Network Every beacon contains a TIM (traffic indicator message). The station notices from the TIM that it has frame buffered at the AP. It then sends a PS-Poll to the AP. The AP then sends the buffered frame to the station. 14- 23

24. Power Save in Infrastructure Network The AP will not send a broadcast or multicast frame if there is one station in the power-save mode. DTIM (delivery TIM) is a special TIM where the AP will send broadcast or multicast frames regardless of the power-save states of stations. 14- 24

25. Power Save in Ad Hoc Network 14- 25

26. Power Save in Ad Hoc Network Similar to the power save mechanism in infrastructure mode An ad hoc station needs to buffer a frame if the frame’s destination station is in the power-save mode. (how to know? Interesting research issue!) After every beacon and within the ATIM (ad hoc TIM) window, the ad hoc source station sends a ATIM frame to the destination station to notify it. Because every ad hoc station is required to wake up at each beacon time (actually can wake up at every N’th beacon) and last UP for ATIM window, it will get this notification. Then the ad hoc source station can send a frame to the destination station directly. 14- 26

27. Two Points We see that in order for every station to wake up at the same time to receive beacons (its TIM) and ATIM, the clocks of these stations need to be synchronized. –This is why in either ad hoc or infrastructure networks, beacon transmission is still needed to synchronize the times of stations. –Since frequency hopping is no longer used for the 11 Mbps PHY, power management seems to be the only main reason why stations’ clocks need to be precisely synchronized. When to start sleeping and wake up, how long to sleep and wake up? How should a station do so that it can save the most power while remaining very responsive is still an interesting research problem. 14- 27

28. (Roaming) Switch to a Different AP A station needs to switch to a different AP when it detects that the quality of the communication channel between it and the current AP is becoming worse. At this time, it should start scanning other available APs. How to determine when to start and stop searching for a better AP is interesting, especially when the station is at the interception of overlapping cells. One technique is to use a high and low watermark thresholds to avoid too much association and disassociation request overhead. (I.e., we should use damping to avoid flapping) 14- 28

29. Control Frame Format Duration = Txtime(data) + Txtime(CTS) + Txtime(ACK) + 3 * SIFS 14- 29

30. Control Frame Format Duration = the duration in the just received RTS frame – Txtime(CTS) – SIFS. 14- 30

31. Control Frame Format Duration = 0 if the “more fragment” bit is 0 in the just received data frame. Otherwise, duration = the duration in the just received fragment – Txtime(Ack) – SIFS. 14- 31

32. Control Frame Format BSSID is the address of the AP. TA is the transmitter’s address. AID is the association ID assigned to this station. 14- 32

33. Control Frame Format The BSSID is the address of the AP. RA is the broadcast address. Duration is 0. 14- 33

34. Control Frame Format Piggyback CF-END and CF-ACK together. The BSSID is the address of the AP. RA is the broadcast address. Duration is 0. 14- 34

35. Data Frame Format 14- 35

36. Management Frame Format BSSID: the ID of the BSS that this frame belongs to. 14- 36

37. Management Frame Format Beacon 14- 37

38. Management Frame Format SSID: specify which BSS the station wants to join. Listen interval: indicate to the AP how often this station will wake up to listen to its beacons. (unit is beacons) Capability: indicate to the AP whether this station can accept PCF. 14- 38

39. Management Frame Format Capability: indicate to the station the capability of the AP. (e.g., whether the AP can do PCF or power management) 14- 39

40. Management Frame Format A station sends this request to a new AP. Current AP address: the new AP can use this information to ask the old AP to forward the station’s data frames to it. 14- 40

41. Management Frame Format 14- 41

42. Management Frame Format Used in active scanning. SSID: can be a broadcast address meaning that the station can accept any AP. 14- 42

43. Management Frame Format Disassociation: it is just a notification. An approval is not needed. Used when a station switches to a different AP, or an AP decides to power down. 14- 43

44. Management Frame Format The results of doing active scanning. Like the beacon format. 14- 44

45. Management Frame Format Because a station can be associated with an AP, it must be authenticated first. One authentication algorithm is OPEN, which mean “always succeed”. 14- 45

46. Management Frame Format 14- 46

47. Management Frame Format ESS = 1 IBSS = 0 : Infrastructure mode ESS = 0 IBSS = 1 : Ad hoc mode 14- 47

48. (Element ID, Length, Information) is a general format. Information Element 14- 48

49. Information Element Current channel: The station will use this information to set its channel to join an AP. 14- 49

50. Information Element CFP count: indicate how many DTIM appear before the next CFP starts. CFP period: indicate the number of DTIM interval between the start of CFPs. CFP Maxduration: indicate the maximum duration of the CFP that may be generated by this PCF. CFP DurRemaining: indicate the time remaining in the current CFP. 14- 50

51. Information Element DTIM count: indicate how many beacons appear before the next DTIM. DTIM period: indicate the number of beacon intervals between successive DTIM. Partial virtual bitmap: each bit indicates whether there is a buffered frame at the AP for a particular station. 14- 51

52. Conclusions 802.11 wireless LAN is becoming more and more popular. When its cost comes down like Ethernet, it will get even wider uses. Thus, knowing the 802.11 standard in detail will be good for you for a long time. 14- 52