This paper considers the random early detection (RED) active queue management algorithm in TCP/IP networks. A stochastic framework for modeling the performance of the RED algorithm is presented. We introduce a two-dimensional second-order discrete-time Markov chain to model the feedback effect of packet dropping on the incoming traffic. The mean system occupancy, packet drop probability, and system throughput are computed from the analytical model for both marking and dropping policies. The effects of RED parameters on a selected set of performance metrics are studied.

A stochastic framework for modeling the performance of the RED algorithm is presented. A two-dimensional second-order discrete-time Markov chain is introduced to model the feedback effect of packet dropping on the incoming traffic. The mean system occupancy, packet drop probability, and system throughput are computed from the analytical model. The effects of RED parameters on a selected set of performance metrics are studied.

We consider routing and wavelength assignment in wavelength-routed all-optical networks (WAN) with circuit switching. The conventional approaches to address this issue consider the two aspects of the problem disjointly by first finding a route from a predetermined set of candidate paths and then searching for an appropriate wavelength assignment. We adopt a more general approach in which we consider all paths between a source-destination (s-d) pair and incorporate network state information into the routing decision. This approach performs routing and wavelength assignment jointly and adaptively, and outperforms fixed routing techniques. We present adaptive routing and wavelength assignment algorithms and evaluate their blocking performance. We obtain an analytical technique to compute approximate blocking probabilities for networks employing fixed and alternate routing. The analysis can also accommodate networks with multiple fibers per link. The blocking performance of the proposed adaptive routing algorithms are compared along with their computational complexity.

This paper studies the effects of tuning delay of transmitters in packet-based optical broadcast networks. We consider scheduling of random traffic with tunable transmitters and fixed-tuned receivers and obtain the degradation imposed by tuning delay using several performance criteria, such as schedule completion time, average packet delay, and session blocking rates. We show that for off-line scheduling the effects of tuning delay are small even if the tuning time is as large as the packet duration. We provide a lower bound to the expected completion time of any off-line schedule with an arbitrary number of wavelengths. We then describe a near-optimal schedule which is based on the principle of having idle transmitters tune to wavelengths just-in-time to start their transmissions. Stability and capacity issues in the transmission of real-time traffic are considered and a queueing-theoretic analysis of average packet delay is given. The packet delay is found to be insensitive to tuning delay under near-optimal transmission scheduling. Finally we extend the model to connection-oriented networks and evaluate the session blocking performance for scheduled circuit connections.

We consider the effects of tuning delay in optical broadcast networks. We show that for off-line scheduling these effects are small even if the tuning time is as large as the packet duration. In particular, we consider scheduling of random traffic with tunable transmitters and fixed-tuned receivers. We provide a lower bound to the completion time of any off-line schedule with an arbitrary number of wavelengths. We then describe a near-optimal schedule which is based on the principle of having idle transmitters tune to wavelengths just-in-time to start their transmissions. Stability and capacity issues in the transmission of real-time traffic are considered. We show that the scheduling problem admits a single stable equilibrium point, and point out how the traffic capacity of a broadcast network can be reached. We also consider the implications in connection-oriented networks.