Wireless Network Management Issues: Current 802.11 Limitations July 2004 IEEE 802.11-04/0000r0 September 2004 Wireless Network Management Issues: Current 802.11 Limitations Joe Kwak, Marian Rudolf / InterDigital Kwak, Rudolf <include authors here>

July 2004 IEEE 802.11-04/0000r0 September 2004 802.11 Today 802.11a,b,g Association/Re-association always initiated by non-AP STA But Disassociation initiated by any STA (AP or not) Minimal signaling to help STAs set their parameters Supported rates, capability info No spectrum or transmit power management 802.11h Applies to 5 GHz band Spectrum management (e.g. channel switch) Transmission power constraint 802.11k in ballot Measurement reports Several new measurements Kwak, Rudolf <include authors here>

Need for Managed Frequency Relocation July 2004 IEEE 802.11-04/0000r0 September 2004 Need for Managed Frequency Relocation Kwak, Rudolf <include authors here>

July 2004 IEEE 802.11-04/0000r0 September 2004 What we cannot do today Interference in the 2.4 GHz band can be a problem E.g. microwave oven, another WLAN system, etc. Could be mitigated by switching to a less affected channel Channel switch Channel 11 – severe interference Channel 1 – less interference The AP can’t switch channel without disruption for STAs Channel switch notification not available in 2.4 GHz band, limited in 5 GHz. STAs have to re-associate to AP (or other AP) following the channel change “Hard” service interruption not acceptable for voice or streaming users Kwak, Rudolf <include authors here>

July 2004 IEEE 802.11-04/0000r0 September 2004 What can we do? DFS service of 802.11h applied to 2.4 GHz could be a starting point. Some of the limitations of 802.11h should be addressed: Fact that channel switch announcements are broadcasted means AP has no way of knowing if the associated STA(s) heard the announcement. Even if the STA heard the announcement, the AP does not know if the STA can/will change channel or when it did. This leads to the AP repeatedly transmitting buffered packets to STA that have not followed on the new channel. Some sort of hand-shaking mechanism is desirable Kwak, Rudolf <include authors here>

Need for Managed Load Balancing July 2004 IEEE 802.11-04/0000r0 September 2004 Need for Managed Load Balancing Kwak, Rudolf <include authors here>

Motives for Load Balancing July 2004 IEEE 802.11-04/0000r0 September 2004 Motives for Load Balancing Traffic in WLAN environments is characterized as being particularly non-uniform and bursty Dynamic load balancing allows for a better distribution and more efficient use of the radio resources The load balancing can be driven by: The STA The AP Optimal performance achieved when load balancing is performed in a coherent and cohesive manner Kwak, Rudolf <include authors here>

Current situation STA-driven load balancing: AP-driven load balancing: July 2004 IEEE 802.11-04/0000r0 September 2004 Current situation STA-driven load balancing: Some STAs have it, others don’t Each manufacturer uses its own algorithms No cohesion behind the decision-making Hard to execute without compromising QoS (e.g. real-time services) Optimal performance require new signaling  proprietary solutions AP-driven load balancing: Allows for a more coherent decision-making AP can only disassociate STA, STA must initiate reassociation Hard to satisfy QoS of real-time users Optimal performance requires new signaling  proprietary solutions Need for standardization - explicitly outside of 802.11r WNM can envisage various levels of control that AP could have in load balancing Kwak, Rudolf <include authors here>

Various levels of control of AP for Load Balancing July 2004 IEEE 802.11-04/0000r0 September 2004 Various levels of control of AP for Load Balancing Low AP sending special beacons on other channels to advertise itself when adjacent BSSs are more loaded Medium Lightly-loaded AP sending association requests to STAs served by highly loaded APs High Serving AP instructing STA to associate to another AP (may involve prior hand-shaking between APs) Kwak, Rudolf <include authors here>

September 2004 CONCLUSIONS As work begins in WNM, scenarios describing current limitations and desired future functions will emerge. WNM needs to address management of seamless BSS Frequency Relocation to mitigate against transient sources of interference. WNM needs to address management of seamless Load Balancing to adapt to dynamic traffic loads. Kwak, Rudolf