Improving TCI/IP performance over wireless networks

Abstract

TCP is a reliable transport protocol tuned to perform well in traditional networks made up of links with low bit-error rates. Networks with higher bit-error rates, such as those with wireless links and mobile hosts, violate many of the assumptions made by TCP, causing degraded end-to-end performance. In tbis paper, we describe the design and implementation of a simple protocol, called the snoop protocol, that improves TCP performance in wireless networks. The protocol modifies network-layer software mainly at a base station and preserves end-to-end TCP semantics. The main idea of the protocol is to cache packets at the base station and perform local retransmissions across the wireless link. We have implemented the snoop protocol on a wireless testbed consisting of IBM ThinkPad laptops and i486 base stations communicating over an AT&T Wavelan. Our experiments show that it is significantly more robust at dealing with unreliable wireless links as compared to normal TCP; we have achieved throughput speedups of up to 20 times over regular TCP in our experiments with the protocol. Intrcdu.ctionRecent activity in mobile computing and wireless networks strongly indicates that mobile computers and their wireless communiication links will be an integral part of future internetworks.. Communication over wireless links is characterized by limited bandwidth, high latencies, high bit-error rates and temporary disconnections that must be dealt with by network protocols and applications. In addition, protocols and applications have to handle user mobility and the handoffs that occur as users move from cell to cell in cellular wireless networks. These handoffs involve transfer of communication state (typically network-level state) from 1. 2 one base station (a router between a wired and wireless network) to another, and typically last anywhere between a few tens to a few hundreds of milliseconds.Reliable transport protocols such as TCP [Pos81, Ste94, Bra891 have been tuned for traditional networks made up of wired links ant3 stationary hosts. TCP performs very well on such networks by adapting to end-to-end delays and packet losses caused by congestion. TCP provides reliability by maintaining a running average of estimated round-trip delay and mean deviation, and by retransmitting any packet whose acknowledgment is not received within four times the deviation from the average. Due to the relatively low bit-error rates over wired networks, all packet losses are correctly assumed to be because of congestion.In the presence of the high error rates and intermittent connectivity characteristic of wireless links, TCP reacts to packet losses as it would in the wired environment: it drops its transmission window size before retransmitting packets, initiates congestion control or avoidance mechanisms (e.g., slow start [Jac88]) and resets its retransmission timer (Karn's Algorithm [KP87]). These measures result in an unnecessary reduction in the link's bandwidth utilizatior:, thereby causing a significant degradation in performance in the ...

Performance improvement in wireless networks using cross-layer ARQ

“…One of the first of such attempts to mask losses on the wireless link from a fixed sender is proposed in [10]: the snoop agent introduced at the base station performs local retransmissions triggered by sniffing duplicate ACKs coming from the wireless receiver.…”

“…In this scenario, mobile nodes are connected to the core network using a set of mobile routers serving as bridges between fixed network infrastructure and mobile devices. Multiple studies show that TCP performance is poor in such environments [10], the main reason of which lies in completely different characteristics between fixed (wired) and wireless parts along the end-to-end data connection [8].…”

Buffer Coding for Reliable Transmissions over Wireless Networks

“…However, due to the high bit-error rates of wireless links and losses induced by interference and mobility, supporting reliable transmissions over wireless networks remains a challenging problem, especially for networking devices with limited resources. To deal with the high loss rate of wireless networks and improve end-to-end throughput, many existing reliable protocols support in-network packet recovery [2][3][4][5]. In these protocols packet caching at the intermediate nodes is used for faster and low-overhead recovery (compared to recovery from source) of lost packets.…”

“…Suppose there is a TCP flow going through the base station. In a traditional protocol supporting local recovery, like Snoop [3], the base station caches packets received, forwards them to the receiver, and retransmits the lost ones upon receiving an ACK indicating packet loss from the receiver. Assume that the wireless link experiences random losses and the packet loss rate is low, say less than 10%.…”

“…As a case-study, we show how XBC can be integrated with the Snoop protocol [3]. We choose Snoop as the baseline because it is well known and readily available for use for both simulations and experiments.…”