Where to begin? For each chapter from the ONT book I read that chapter, read the same topics in the QoS Exam Certification Guide, begin to take notes and then decide what feels lacking in my understanding. From that gut feeling I begin to branch out, starting with Cisco documentation, then trustworthy sources on the net.
WLAN QoS has morphed on me. ONT covers split-mac, WMM and implementation with coverage of 802.11e nearly non-existent. Not only that, but the portions of 802.11 covered in the book are narrow. For instance 802.11 defines two channel access mechanisms, distributed coordination function (DCF) which is covered, but point coordination function (PCF) is not even mentioned. As a result, I’m just going to fill in what makes sense to me as I believe having a basic understanding of WLAN QoS and wireless in general is the purpose of the wireless section anyway.
- 802.11e — QoS extensions for WLANs.
- Wi-Fi Multimedia (WMM) — A subset of 802.11e with four access categories; voice, video, best effort and background. The WCS breaks them out as platinum, gold, silver and bronze.
- Contention Window — CWmin and CWmax — Define the the range of the contention window, the back off timer in CSMA/CA. A client with a voice CWmin waits less time to send than a client in the best effort access category that has a larger CWmin.
- Arbitrary Inter-Frame Space Number (AIFSN) — Controls the idle time, after which a client may transmit.
- Transmission Opportunity (TXOP) — The time interval, start and maximum duration, a client holds the channel and is able to send data.
Wired clients use carrier sense multiple access with collision detection (CSMA/CD) to determine whether they can transmit, however, one of its cousins is token ring which is rarely used today. In the same manner wireless access has two channel access mechanisms, distributed coordination function (DCF) and point coordination function (PCF). Like token ring, PCF is rarely used. DCF uses carriers sense multiple access with collision avoidance (CSMA/CA) which on it’s own provides best effort delivery.
- Distributed Coordination Function (DCF) is composed of two main components:
- Interframe spaces — SIFS, PIFS and DIFS control access to the channel.
- Contention window — The random backoff timer.
- Interframe spacing is a portion of the time a client or AP waits before sending frames on the channel. It is independent of the backoff time or contention window.
- Short Interframe Space (SIFS) — 802.11 management frames and those not expecting contention because they are port of a sequence of frames.
- Point Coordination Function Interframe Space (PIFS) — Used by an AP to decide when to send, shorter than DIFS giving the AP priority over data, but longer than SIFS giving flows priority. SIFS < PIFS < DIFS
- Distributed Coordination Function Interframe Space (DIFS) — When a client begins to send new data it waits DIFS, checking for a clear channel before sending.
802.11e and Wi-Fi multimedia (WMM) provide QoS for wireless networks. WMM was implemented by the Wif-Fi Alliance before 802.11e was ratified. I think of it as wireless QoS light as it implements a subset of 802.11e. 802.11e and WMM use Enhanced DCF (EDCF) to provide proportional back-off window sizes for each class.
- 802.11e defines enhancements to 802.11 Medium Access Control (MAC) to provide QoS features by using features from both PCF and DCF in what is called Hybrid Coordination Function (HCF).
- 802.11e provides eight priority levels, 0 through 7.
- Different acknowledgement rules provide greater efficiency in being able to send data.
- Piggybacking allows data to be sent with poll requests and ACKs improving network performance.
- Allows for contention free bursts, clients and APs are able to send several frames without contention.
- HCF has two modes of operation:
- Enhanced Distribution Coordinate Access (EDCA)
- HCF Controlled Channel Access (HCCA)
- Enhanced Distribution Coordinate Access (EDCA) is an extension to DCF that uses contention based access while providing prioritized access to the channel.
- EDCA has four key components:
- CWmin with higher priority assigned a shorter CWmin.
- TXOP limit specifies the maximum duration a client can transmit, makes channel access more efficient.
- Arbitration Inter-Frame Space (AIFS) specifies additional time between when a channel goes idle and the client starts to send. Each access class is assigned a different AIFS to further differentiate QoS.
- HCF Controlled Channel Access (HCCA) is polling based and is uses a coordinator to centrally manage access.
- HCCA can poll clients during the contention period.
- Supports scheduling of packets based on traffic flow requirements.
- The coordinator has the highest priority access.
802.11e priorities can be mapped to WMM access categories for backward compatibility:
|Voice (Platinum)||6 or 7|
|Video (Gold)||4 or 5|
|Best-Effort the Default (Silver)||0 or 3|
|Background (Bronze)||1 or 2|
The image below is my recreation of a figure out of the Wi-Fi Alliance documentation below. It explains the interaction between AIFSN, CWmin/CWmax and it’s affect on WMM. In essence a voice client waits less time before trying to retransmit than a lower access category and will therefore have a better chance at sending data, it is not a strict priority system.
Split MAC Architecture
Centralizes wireless LAN configuration and control onto the wireless lan controller (WLC). Access points are lightweight and cannot act independently of a controller. The wireless LAN controller manages the access point configurations and firmware. The access points handle only real-time MAC functionality, leaving all the non-real-time MAC functionality to be processed by the wireless LAN controller.
- WLC manages:
- Channel assignment
- Association, disassociation and reassociation
- 802.11e and WMM resource reservation
- Transmit power optimization
- Self-healing wireless coverage
- Dynamic client load balancing
- LWAP manages:
- Transmission of beacon frames
- Probe transmission and response
- 802.11e and WMM scheduling and queuing
- Buffering frames for clients
- Monitoring each radio channel for noise and interference
Light weight access point protocol (LWAPP)
Light weight access point protocol (LWAPP) is used to communicate between the WLC and the APs. Wireless LAN client data packets are encapsulated in LWAPP between the access point and the wireless LAN controller. When a wireless LAN client sends a packet, it is received by the access point, decrypted if necessary, encapsulated with an LWAPP header and forwarded to the controller. At the controller, the LWAPP header is removed and the frame switched from the controller onto a vitrual LAN (VLAN) in the switching infrastructure.
There are two types of LWAPP traffic, control messages and client data.
- LWAPP control messages:
- Used to configure the LAP and manage its operation.
- Authenticated and encrypted, the LAP is securely controlled by only the WLC.
- Classified automatically with a DSCP of CS6.
- Identified by UDP port 12223
- LWAPP data:
- Packets to and from wireless clients associated with the LAP.
- The data is encapsulated within LWAPP but is not encrypted.
- The default classification for WLAN data traffic is best-effort.
- Identified by UPD port 12222
WMM from Wi-Fi.org
Cisco Mobility Design Guide
LWAPP Traffic Study
WLC Deployment Guide
Intel QoS Paper
European Wireless Conference Paper