Network Flow and Power Control

 

Distributed optimization is a natural approach to balance performance, fairness, and constraints in network traffic management.  For route selection and flow partition, optimization has been proposed to match traffic demand with network capacity and topology.  For the transport layer protocol, flow control has been posed as a utility maximization subject to the link bandwidth constraint.  Power control in wireless networks has been posed as an optimization involving signal-to-interference ratio (SIR) and transmission power.  In mobile ad hoc networks, rate and power control may be simultaneously considered in the optimization of throughput subject to the capacity and power constraints.

 

In highly dynamic networks, to achieve more efficient resource utilization and higher system throughput, routing and flow control (and power management in wireless networks) need to be treated in an integrated manner.  Despite the similarity between these network traffic management optimizations (all involve distributed optimization subject to constraints), they are currently treated as isolated problems. Furthermore, the transient and robustness properties of networks, which are critical to the network performance in a dynamic environment, are rarely addressed within the optimization framework.

 

The objective of this project is to develop a unifying optimization-based methodology for data network traffic management with integrated consideration of stability, transient performance, robustness (to disturbances, delays, and uncooperative users), scalability (to network size), and quality of service guarantees.  The main technical tool is passivity, a control-theoretic concept, which we recently introduced to the area of network flow control.  It is an ideal tool for network analysis and design due to its applicability to nonlinear systems and close linkage to optimization.  A system with state x, input u, and output y, is said to be passive if there exists a continuously differentiable ``storage function'' V(x)≥0 such that dV/dt ≤ -W(x)+uTy, for some W(x) )≥0.  This passivity concept is motivated by energy conservation or dissipation in physical systems (uTy can be considered as the power delivered to the system), for example, passive circuits, mechanical structures, and thermal systems.  Passivity has long been used in the stability analysis and control design for nonlinear feedback systems, and has been instrumental in recent breakthroughs in nonlinear control design.  In the context of flow control, we have shown that past stabilizing network congestion control laws in are special cases of our passivity approach.  The passivity approach has also been shown to be applicable to uplink power control in cellular systems. Our current studies indicate that the passivity approach is also suitable for distributed route optimization.  We are also pursuing joint flow, routing, power, and position control in mobile ad hoc networks.

 

Network flow control

 

CDMA power control

 

Publications:

Acknowledgment
This research is supported in part by the RPI Office of Research through an Exploratory Seed Grant.  This work is also supported in part by the Center for Automation Technologies (CAT) under a block grant from the New York State Office of Science, Technology, and Academic Research (NYSTAR).  This work is also supported by the China NSFC two-base project under grant no. 60440420130.

Contact Information:

John T. Wen
Office: CII 8213
Voice: (518)-276-8744
Fax: (518)-276-4897
Email: wen@cat.rpi.edu