We derive a modular fluid-flow network congestion control model based on a law of fundamental nature in networks: the conservation of information. Network elements such as queues, users, and transmission channels and network performance indicators like sending/acknowledgment rates and delays are mathematically modeled by applying this law locally. Our contributions are twofold. First, we introduce a modular metamodel that is sufficiently generic to represent any network topology. The proposed model is composed of building blocks that implement mechanisms ignored by the existing ones, which can be recovered from exact reduction or approximation of this new model. Second, we provide a novel classification of previously proposed models in the literature and show that they are often not capable of capturing the transient behavior of the network precisely. Numerical results obtained from packet-level simulations demonstrate the accuracy of the proposed model.

In a cellular network it is necessary to make a trade-off between congestion and Quality of Service (QoS) of the users. To ensure good system performance in uplink of a WCDMA network, control loops are used to adapt to changing radio and interference conditions. A fast inner power control loop updates the transmission powers of the mobiles based on measurements related to QoS, which are compared to a reference value. The reference value is set dynamically by a slower outer rate control loop that uses measurements of congestion. There is a time-scale difference between the loops, but joint dynamics cannot be neglected. In this paper we derive a high order system model with distributed feedback control based on locally measurable information. We model the time-scale difference between the loops and include this in the model. We give sufficient conditions for stability using control theoretic methods and validate the model and results by simulations.

Mobile ad hoc networking allows nodes to form temporary networks and communicate beyond the transmitter range by supporting multihop communication through IP routing. Routing in such networks is often reactive, i.e., performed on-demand, as opposed to Internet routing that is proactive. As ad hoc networks are formed on a temporary basis, any IP address should be allowed to appear in an ad hoc network. This paper presents MIPMANET, a solution for connecting an ad hoc network, in which on-demand routing is used, to the Internet. MIPMANET provides Internet access by using mobile IP with foreign agent care-of addresses and reverse tunneling. This allows nodes to enjoy the mobility services of mobile IP while at the same time the requirements on the ad hoc routing protocol are kept to a minimum. Simulations of MIPMANET have been performed in the Network Simulator 2. The ad hoc on-demand distance vector (AODV) routing protocol has been used for routing within the ad hoc network. These simulations show that the ability to choose the closest access point to the Internet is worth extra work, as less traffic is generated in the network resulting in lower delays and fewer dropped packets.