ICWMMN2006 Proceedings
IEEE 802.11 WLAN load balancing using adaptive antennas and cooperative controls Yapeng Wang, Laurie Cuthbert and Athen Ma MPI-QMUL Information Systems Research Centre, Macau Polytechnic Institute Macau SAR, P. R. China E-mail:
[email protected]
Abstract This paper extends our recent researches on using intelligent agents to improve 802.11 WLA N systems performance. Agent-controlled semi-smart antennas and multi-agent systems are proposed. The use of agent negotiation and cooperatively change antennas' radiation patterns can significantly improve overall data rates for WLA N users , especially for geographically non-uniform distributed users (so called "hot spot" traffic). Distributed agents can negotiate with each other resulting mutual changing radio patterns to re-balance traffic at hot spot area. The simulation has taken multi-path effects into cons ideration, as most WLAN systems are deployed in in-door environment. Simulation results show significant overall system performance improvement over traditional configurations. A network planning tool is used to create path loss models based on building structures and floor plans . The design and building of prototype semi-smart antennas and their control units are also discussed. Keywords: balancing
Intelligent Agent, Smart Antennas, WLAN load
1 Introduction The IEEE 802.11 based Wireless Local Area Networks (WLAN) system has been rapidly developed and occupy the market in recent years. The original 802.11 standard specifies a common Med ium Access Control (MAC) layer and three Physical (PHY) layers [1]. Two of these PHYs facilitate communications in the 2.40Hz Industrial Scientific and Medical (ISM) band s using Direct Sequence (OS) and Frequency Hopped (FH) Spread Spectrum techniques. Data rates of up to 2 Mbits /s are facilitated by each of the PHYs. The 802.11 standard has subsequently been expanded considerably with several PHY extensions the data rate can support up to 54 Mbits /s. In a multi-access WLAN system, the MAC layer is the main clement for determining the efficiency in sharing the communication bandwidth of the wireless channel. The DCF is mandatory and most common used for 802.11 devices and is based on a Carrier Sense Multiple Access with Collision Avoidance (CSMNCA), As all stations in an 802.11 WLAN implement the same carrier sense and back off mechanism, all stations fairly share the communication channel in statistics for a long period of time . In large area public deployment of 802.11 WLAN systems, multiple Access Points (APs) are used and geographically distributed in an overlapping manner (similar to cellular systems). Each AP normally coordinates one air channel and covers a certain area offering all stations (ST A) accessing the AP . When more STAs accessing a same AP with high application demand (such as downloading),
These WLAN users will experience lower data throughput due to the congested air channel. Based on experience of load balancing for COMA systems [2], we propose a dynamic, distributed intelligent agent controlled WLAN system to cope with congestions and performing automatic load balancing with the help of novel semismart sectored antennas for APs .
2 The Proposed Semi-smart Sectored Antenna for Access Point For large scale WLAN systems, the "cell" concept (in cellular networks) can still be used to represent the Access Point (AP) layout and multiple APs can cover a wider area seamlessly to support crossAP mobility. The most common antennas used in WLAN AP are omni-directional diverse antennas, which apply a circular shape coverage area. In order to achieve more controllable radio coverage patterns, we propose 4-sector semi-smart array antenna for APs . Each sector has multiple antenna clements and individual power control in order to achieve directional and controllable pattern, thus the combination of four sector antennas will give an AP the ability to produce flexible radio coverage patterns (Figure I) . The four sector antenna patterns add to give the overall cell coverage pattern. Design of a four-sector antenna array is a compromise between performance and cost. More sectors will increase the flexibility of coverage shaping but has disadvantages: a) More RF power controllers are needed. b) Each sector needs a narrower antenna pattern, so that each sector antenna needs to be physically bigger. In Figure I, A WLAN access point is linked to our semi-smart antenna system . The external antenna connection is first connecting to a 4 way bi-d irectional signal splitter/combiner in order to feed the 4 separate antennas. Then it passes through a digital step attenuator. All attenuators are connected to a control unit (firstly it will be a PC computer; single chip computer embedded in access point can be used for large scale deployments). The digital step attenuator used here 1 to control the signal power to be feed into the panel antennas. It is a flexible 6 Bit digital parallel controlled unit that can have maximum 3 1.5 dB attenuation with a minimum 0.5 dB step. The four antenna used here are commercial product", each of the four panel can produce a horizontal 90 0 degree beam width , so four of the antenna will cover a whole 360 0 (Figure 2)
1 The digital attenuator is a product of Mini-Circuits, Internet: http ://www.minicircuits.com 2 2.4 OHz panel antenna manufactured by HyperLink Technologies, Internet: http://www.hyperlinktech.com
J
