Wireless Technology: Issues on QoS Stability, Sensitivity, Flow & Congestion Control in Wireless Broadband Networks

University Of California-Irvine, Irvine CA

Investigators

Abstract

There has been a lot of research work on Wireless ATM in the past 5 years; most of the research works are geared towards mobility and connection management. The former is concerned with location and handoff management while the latter is concerned with signaling for call setup and teardown. What are remaining are still some very open issues that demand answers. In supporting QoS in Wireless ATM, the variation of QoS availability over space and time is different from one wireless cell to another. Mobile calls that are established at call setup may longer to able to maintain their QoS demands if the mobile host migrates from cell to cell, unless pre-allocation of network resources are made in advance (however, this is very inefficient). The trend so far has been towards an adaptive end-to-end Mobile QoS framework for Wireless ATM networks. Although this proposal addresses mobile QoS for Wireless ATM, it is not another duplication of existing efforts. Instead, we are investigating this area of work from another dimension - applying the know-how of control and network theory and protocols. It shall address challenging research issues at the network and transport layers and hence this proposal demands vertically integrated research. Most end-to-end QoS adaptation framework attempts to ensure the QoS (normally the bandwidth and delay requirements) over both wired and wireless links are consistent after a handoff. However, such handoffs can occur on the two remote ends of the mobile-to-mobile connection. If the adaptation protocols are not well designed, the WATM can experience instability problems. The fact that a mobile host or the network triggers an adaptation is a sign of pro-active reaction. This is called sensitivity. A highly sensitive system can also cause substantial signaling overhead and result in network congestion and instability. The instability refers to a regenerative phenomenon where another adaptation process is triggered even before the current one is completed. The innovative approach is to apply control theory and to derive new network algorithms. We plan to introduce a set of damping procedures to control sensitivity and instability associated with continuous and transient oscillation in QoS adaptation. Our plan is to derive these adaptation damping procedures and to implement them on a WATM simulator/testbed to evaluate the feasibility of the approach and the resultant stability and sensitivity of the system under different traffic load, degree of mobility and number of users. This area of this research proposal, therefore, enhances QoS support and more importantly, the reliability, availability and performance of future broadband wireless network systems. It views the understanding of underlying dynamics associated with QoS adaptation paramount to building high performance and reliable wireless broadband systems. The second part of the research concerns with ABR and VBR flow control in WATM networks. While there has been much on-going research associated with ATM flow and congestion control, little is devoted to wireless flow and congestion control in WATM networks. This proposal reveals the impact of the dynamics associated with SS (Stability and Sensitivity) in QoS adaptation on ABR/VBR flow control. This is a very genuine problem since current rate-based flow control relies on the use of feedback cells and these cells can be lost as a result of host mobility. The innovative approach has two components: (a) the separation of rate allocation and signaling in a vertically integrated design, (b) the optimization of flow control functions in both space and time. It intends to develop and validate a theory of protocols that enables max-min optimization, enhancing stability and minimizing sensitivity to an appropriate level. Protocols designed based on this theory will be amenable to rigorous analysis and proofs of behavior.