Communications Network Research Institute

WLAN Radio Resource Management

Mark Davis

Increasingly cellular network operators are looking to mobile data offload solutions to reduce congestion on 3G cellular networks arising from the explosion in the demand for video services. Mobile data offload involves the use of complementary network technologies (i.e. Wi-Fi hotspots) for delivering data originally targeted for cellular networks. However, the use of Wi-Fi (i.e. IEEE 802.11 WLAN) networks to deliver real-time services such as telephony (e.g. Skype Wi-Fi phone) and video streaming presents major challenges in terms of how the radio resources (i.e. the bandwidth) of the network is managed.

An essential prerequisite for implementing any radio resource management (RRM) scheme is the ability to measure the network bandwidth utilization and availability or to put it another way:

Unless you can measure it, you cannot control it

Estimating the available bandwidth or the capacity of a Wi-Fi network is problematic owing to the nature of the medium access control (L2/MAC) mechanism used which is based upon random access whereby wireless nodes must compete with one another in order to win transmission opportunities for their packets.

A consequence of having to compete for transmission opportunities is that the capacity of the network is not fixed and instead varies with the level and the nature of the offered load. Furthermore, different wireless nodes will perceive different capacities depending on what they are doing and on what their neighbouring nodes are doing. This makes the task of estimating the capacity difficult to achieve and is one of the main obstacles to implementing effective RRM schemes on Wi-Fi networks.

The CNRI has developed a mathematical framework for modelling how the bandwidth of a WLAN is shared between competing wireless nodes. This framework is based upon the concept of MAC bandwidth components which describe how the MAC mechanism employed in the IEEE 802.11 standard controls the bandwidth of the wireless medium. This radio resource framework is used to provide for measurements of the utilization and availability of bandwidth experienced by all of the competing network nodes.

The CNRI has developed a WLAN Resource Monitor (WRM) application that is capable of measuring both the availability and utilization of network bandwidth in real time and on a per node basis. It operates non-intrusively by passively monitoring the wireless transmissions on the medium (i.e. it sniffs the medium) and therefore does not in any way interfere with the normal flow of traffic on the network. Moreover, it does not require WEP/WPA security keys to operate and therefore does not pose any security risks when deployed on encrypted networks. The WRM application differs from other WLAN Analyzer and Wi-Fi sniffer tools in that it specifically addresses operation at the L2/MAC layer which is where the network bandwidth is shared out between the competing nodes. Consequently, this application can provide the type of critical network bandwidth measurements required for effective radio resource management.

Our experiments with this application in the areas of VoIP telephony and Video Streaming on WLANs have shown it to be an important tool for WLAN performance analysis. We have also incorporated it into a novel routing protocol for mesh networks based upon using passive measurements of the link quality to support the path selection mechanism.

A new generation of Wi-Fi equipment has emerged that provides additional support mechanisms for delivering quality of service (QoS) under the IEEE 802.11e MAC QoS Enhancement standard (or Wi-Fi WMM Specification). With Wi-Fi WMM compatible equipment it is now possible to explicitly prioritize real-time services by making it easier for their packets to win transmission opportunities. While this has the benefit of enhancing the QoS experience of the users of these services, it further complicates the problem of estimating the available bandwidth as the nodes no longer compete with each other on an equal basis. Moreover, there will be additional internal competition between the different traffic classes within a wireless node.

The CNRI has also developed a measurement based WLAN Resource Controller (WRC) application that automatically manages the bandwidth of a WLAN network by leveraging the additional control functionalities of Wi-Fi WMM compatible equipment. This application allows the user to allocate a bandwidth capacity to a given traffic class. The WRC adaptively responds to the changing network conditions to ensure that the traffic class continues to experience the bandwidth allocated to it. This WRC application has been used to provision bandwidth to a video stream delivered over a WLAN to ensure that the end user always experiences an acceptable level of quality.