Existing works have approached the problem of reliable transport in ad-hoc networks by proposing mechanisms to improve TCP's performance over such networks. In this paper we show through detailed arguments and simulations that several of the design elements in TCP are fundamentally inappropriate for the unique characteristics of ad-hoc networks. Given that ad-hoc networks are typically stand-alone, we approach the problem of reliable transport from the perspective that it is justifiable to develop an entirely new transport protocol that is not a variant of TCP. Toward this end, we present a new reliable transport layer protocol for ad-hoc networks called ATP (ad-hoc transport protocol). We show through ns2 based simulations that ATP outperforms both default TCP and TCP-ELFN.

We present a new implementation of TCP that is better suited to today's Internet than TCP Reno or Tahoe. Our implementation of TCP, which we call TCP Santa Cruz, is designed to work with path asymmetries, out-of-order packet delivery, and networks with lossy links, limited bandwidth and dynamic changes in delay. The new congestion-control and error-recovery mechanisms in TCP Santa Cruz are based on: using estimates of delay along the forward path, rather than the round-trip delay; reaching a target operating point for the number of packets in the bottleneck of the connection, without congesting the network; and making resilient use of any acknowledgments received over a window, rather than increasing the congestion window by counting the number of returned acknowledgments. We compare TCP Santa Cruz with the Reno and Vegas implementations using the ns2 simulator. The simulation experiments show that TCP Santa Cruz achieves significantly higher throughput, smaller delays, and smaller del...

Wireless access networks in the form of wireless local area networks, home networks, and cellular networks are becoming an integral part of the Internet. Unlike wired networks, random packet loss due to bit errors is not negligible in wireless networks, and this causes significant performance degradation of transmission control protocol (TCP). We propose and study a novel end-to-end congestion control mechanism called TCP Veno that is simple and effective for dealing with random packet loss. A key ingredient of Veno is that it monitors the network congestion level and uses that information to decide whether packet losses are likely to be due to congestion or random bit errors. Specifically: 1) it refines the multiplicative decrease algorithm of TCP Reno---the most widely deployed TCP version in practice---by adjusting the slow-start threshold according to the perceived network congestion level rather than a fixed drop factor and 2) it refines the linear increase algorithm so that the connection can stay longer in an operating region in which the network bandwidth is fully utilized. Based on extensive network testbed experiments and live Internet measurements, we show that Veno can achieve significant throughput improvements without adversely affecting other concurrent TCP connections, including other concurrent Reno connections. In typical wireless access networks with 1% random packet loss rate, throughput improvement of up to 80% can be demonstrated. A salient feature of Veno is that it modifies only the sender-side protocol of Reno without changing the receiver-side protocol stack.

The Internet has evolved the last decade, reaching a larger number of users, and encompassing several new technologies. New classes of hosts such as mobile devices are gaining popularity, while the transmission media become more heterogeneous. Wireless networks exhibit different characteristics than wired ones. Mobile hosts have different needs and limitations than desktop computers. TCP has served well the wired Internet for almost 20 years but is not ready for wired-cum-wireless environments. In this paper we present the challenges that have to be met in order to provide reliable transport services to all hosts regardless of the type of network connectivity used. We survey the recently proposed solutions and evaluate them with respect to a wired-cum-wireless environment.

We present the transport unaware link improvement protocol (TULIP), which dramatically improves the performance of TCP over lossy wireless links, without competing with or modifying the transport- or network-layer protocols. TULIP is tailored for the half-duplex radio links available with today's commercial radios and provides a MAC acceleration feature applicable to collision-avoidance MAC protocols (e.g., IEEE 802.11) to improve throughput. TULIP's timers rely on a maximum propagation delay over the link, rather than performing a round-trip time estimate of the channel delay. The protocol does not require a base station and keeps no TCP state. TULIP is exceptionally robust when bit error rates are high; it maintains high goodput, i.e., only those packets which are in fact dropped on the wireless link are retransmitted and then only when necessary. The performance of TULIP is compared against the performance of the Snoop protocol (a TCP-aware approach) and TCP without link-level retra...

Recent research has focussed on the problems associated with TCP performance in the presence of wireless links and ways to improve its performance. We present an extension to TCP Santa Cruz which improves TCP performance over lossy wireless links. TCP has no mechanism to differentiate random losses on the wireless link from congestion, and therefore treats all losses as congestive. We present a simple method in which our protocol is able to differentiate these random losses, thereby avoiding the rate-halving approach taken by standard TCP whenever any loss is detected. We compare the performance of our protocol against TCP Reno and demonstrate higher throughput and lower end-to-end delay with our approach.

In this paper we focus on self-contention – contention between packets of the same transport layer connection along the path from source to destination. We observe that selfcontention plays an important role in degrading TCP performance in multi-hop wireless networks and that the use of the popular IEEE 802.11 MAC protocol exacerbates selfcontention. We propose and study two MAC-layer approaches to alleviate self-contention. The first approach, called quickexchange (QE), is designed with the intent of reducing the effects of inter-flow self-contention (e.g. between packets of the same connection traveling in opposite directions). The design of our second mechanism, called fast-forward (FF), is geared towards decreasing intra-flow self-contention (e.g. between packets of the same connection traveling in the same direction). We simulate and study our proposed schemes and observe that quick-exchange consistently improves network aggregate goodput (by as much as 20 % in string topologies, 15 % in random static scenarios, and 10 % in random mobile scenarios). In contrast to our expectations, fast-forward causes sporadic and often negative effects on goodput for TCP connections. Upon investigation we find that while the MAC is, in some respect, operating more efficiently, as demonstrated by improved UDP throughput; interactions with TCPs congestion control mechanism cause the goodput to degrade. We analyze various effects that cause the respective behaviors with QE and FF in detail.

This article surveys wireless Internet technologies whose goals are to enhance networking performance.

As wired computer networks support increasingly sophisticated applications and wireless local area networks become ubiquitous and fast, it is more natural for users to seek a "wireless Internet" experience that is qualitatively the same as that provided by the wired Internet. However, wireless LANs pose two fundamental challenges to this vision. Harsh and dynamic error environments challenge end-to-end adaptation at the transport and application layers. In addition, dynamic and location-dependent errors challenge the notion of "fair" scheduling of flows sharing a wireless link.

In this paper, we present the application of machine learning techniques to the improvement of the congestion control of TCP in wired/wireless networks. TCP is suboptimal in hybrid wired/wireless networks because it reacts in the same way to losses due to congestion and losses due to link errors. We thus propose to use machine learning techniques to build automatically a loss classifier from a database obtained by simulations of random network topologies. For this classifier to be useful in this application, it should satisfy both a computational constraint and a time varying constraint on its misclassification rate on congestion losses. Several machine learning algorithms are compared with these two constraints in mind. The best method for this application appears to be decision tree boosting. It outperforms ad hoc classifiers proposed in the networking literature and its combination with TCP improves significantly the bandwidth usage over wireless networks and does not deteriorate the good behaviour of TCP over wired networks. This study thus shows the interest of the application of machine learning techniques for the design of protocol in computer networks. 1