Adaptive Radio Management in 802.11 Networks
|Event Date:||April 9, 2009|
|Speaker Affiliation:||Cofounder, Chief Technology Officer,
Aruba Networks Incorporated
|Sponsor:||CNSIP Area Seminar|
|Contact Name:||Prof James Krogmeier
|Contact Phone:||(764) 494-3530
|Open To:||Acceptable for ECE694A
Adaptive Radio Management 802.11 WLANs operate entirely in two unlicensed bands - 2.4 GHz and 5 GHz. The high proliferation of 802.11 WLANs also create the potential for multiple, uncoordinated wireless networks to overlap in the same physical space. So any 802.11 WLAN solution has to adapt to both the changing RF environment as well as interference from sources that they can coordinate with at the MAC (medium access control) layer such as other 802.11 networks and those that they cannot coordinate with (non-802.11 sources such as Bluetooth, cordless phones, etc.). Roaming is an important benefit that end-users cherish in a wireless network. Roaming creates at least two challenges to a wireless network. The wireless network is subjected to dynamic reconfiguration over periods that are significantly shorter compared to similar changes on a wired network. The wireless network is responsible to migrate user state around the infrastructure such that the end-user perceives a continuous link as they roam between multiple access points (AP). Roaming also creates an uneven distribution of users and traffic in a WLAN that varies with both time and physical space. A wireless LAN has to adapt to this variation in user and traffic load by balancing load across multiple available access points for optimal performance. Aruba Networks created a centralized architecture for large-scale WLANs where user and security state are anchored at a central controller that manages hundreds of access points. Every AP in the WLAN monitors its RF neighborhood and modifies its RF state, if necessary, to achieve the best performance possible. The exchange of information between APs occurs both directly on the wireless link and through the controller where such direct exchange is not immediately possible. In this seminar, we will discuss a subset of the algorithms that were designed to address problems in large-scale WLANs with additional emphasis to problems that occur at locations of high user concentrations (such as a large lecture hall). These algorithms solve problems that include continuous and automatic optimization of the RF environment to optimize coverage while working around interference sources, balancing user and traffic load evenly across APs and channels, and queueing and scheduling of traffic for achieving desired QoS objectives in a network with heterogeneous client mix, and heterogeneous traffic mix.
Keerti Melkote is Aruba's chief technology officer and responsible for product management. Mr. Melkote has 15 years of technology and marketing experience in both enterprise and service provider networking. Prior to founding Aruba Networks, Mr. Melkote held senior management positions at Tahoe Networks, a cellular/3G data networking systems company now part of Nokia, and Shasta Networks, a broadband IP subscriber management systems company now part of Nortel. Mr. Melkote was at Cisco Systems during the LAN switching era in the mid-90s, where he was responsible for the definition and launch of Cisco's high-end enterprise LAN switching systems. Before Cisco, Mr. Melkote was responsible for deploying high-performance LAN and ATM backbone infrastructure at Intel. He holds an MSEE from Purdue University with an emphasis on distributed systems and TCP/IP networking.