Antonio Desimone

Throughput performance of transport-layer protocols over wireless LANs

Considers the performance of transport-layer protocols over networks where one of the links is wireless. A reliable transport protocol is responsible for end-to-end data integrity, and will retransmit lost or errored data. Compared to a link over copper or fiber, a wireless link will have a much higher error rate. Link-layer retransmissions can reduce the error rate on the radio link, but the interaction of link-layer retransmission with end-to-end retransmission can be complicated. The authors have investigated, via analytic, numerical and simulation techniques, the end-to-end effects of link-layer retransmissions in the presence of a reliable end-to-end transport protocol. The results identify the mechanisms affecting end-to-end performance when retransmissions are used to get better error performance on a link. They quantify the increased load on the link due to competing retransmission strategies, and, for a transport protocol modeled on TCP, they identify the region of loss rates where link-layer retransmissions have the undesirable effects of both reducing end-to-end throughput and increasing link utilization in the network segment where bandwidth is the most expensive.<<ETX>>

Controlling alternate routing in general-mesh packet flow networks

High-speed packet networks will begin to support services that need Quality-of-Service (QoS) guarantees. Guaranteeing QoS typically translates to reserving resources for the duration of a call. We propose a state-dependent routing scheme that builds on any base state-independent routing scheme, by routing flows which are blocked on their primary paths (as selected by the state-independent scheme) onto alternate paths in a manner that is guaranteed—under certain Poisson assumptions— to improve on the performance of the base state-independent scheme. Our scheme only requires each node to have state information of those links that are incident on it. Such a scheme is of value when either the base state-independent scheme is already in place and a complete overhaul of the routing algorithm is undesirable, or when the state (reserved flows) of a link changes fast enough that the timely update of state information is infeasible to all possible call-originators. The performance improvements due to our controlled alternate routing scheme are borne out from simulations conducted on a fully-connected 4-node network, as well as on a sparsely-connected 12-node network modeled on the NSFNet T3 Backbone.

Generating burstiness in networks: a simulation study of correlation effects in networks of queues

Packet data networks can have interesting and complicated effects on the traffic streams that share resources in the network. The simulation results reported here indicate that the phenomena of clustering in networks, where the interarrival time arrival of packets in a traffic stream becomes more bursty on passing through a network, is a generic property of networks that carry packets of highly variable size. Clustering, observed as increased variability in the interarrival times of packets, arises even when the network is modeled as a simple open network of queues. Endpoint effects, such as end-to-end flow controls, need not be invoked to explain clustering. At moderate utilizations, traffic streams passing through a series of queues show anomalously large variability in interarrival times, compared to the variability expected from simple decomposition approximations. In fact the anomalous variability can be attributed to a failure of the Kleinrock independence assumption: simulations of the identical network wherein the packets regenerate the service time at each node do not show the anomalous variability.

Performance of two TCP implementations in mobile computing environments

The error recovery algorithm in TCP has been designed for congestion control in wired networks. However, the packet losses in wireless networks are mostly caused by outages due to connection interruptions such as fading, channel preemptions, and handoffs. We investigate the performance of two TCP implementations in mobile computing environments. We consider simplified versions of two TCP implementations (referred to as Protocol I & II). Protocol I is a variation of the TCP Tahoe which uses a fine-grained timer. Protocol II is a simplified version of the TCP Vegas. We first describe an interesting observed phenomena with these two timer-based protocols when the wireless channel is periodically taken away from the end/end connections. Next, we show the robustness of Protocol II and quantify its improvement over Protocol I in terms of achievable goodputs. We also discuss better goodputs can be achieved by setting the RTO differently to account for a higher variability in delays caused by the wireless channel. Finally, we describe how a bigger TCP segment size helps to improve goodputs when these TCP segments are transported over the wireless channel.

Traffic management and priority control in an ATM test-bed

A field grade ATM test bed has been built to understand the role of ATM technology in providing network infrastructure for transporting current and emerging services. Physical, ATM, AAL, and application level testing and measurements are being used to obtain this understanding. One aspect undergoing investigation is the role of cell level traffic controls. In this, we discuss the testing, measurements, and analysis to investigate how ATM traffic controls deployed in switches can be used to manage the resources in a broadband ATM network. Traffic contracts based on UPC and traffic policing at the network edge have been tested to understand how cell-level controls on a per-virtual-circuit basis affect CBR and data applications. The effects of priority controls as implemented in the test-bed have also been studied to understand how diverse applications (voice over CES, bursty data) can share network facilities.