Abstract Wireless links differ from traditional ‘wired’ links in two ways that challenge the existing Internet. On wireless links packet loss or corruption due to transmission errors is not rare, which calls into question the standard Internet assumptions that transmission errors should be corrected by transport‐level protocols at end systems and that end‐to‐end packet loss typically indicates network congestion. Also, the severity and location‐ dependent nature of these errors calls into question the meaning of ‘fair’ scheduling, per‐flow quality of service, and even looser notions such as service level agreements, when applied to wireless links. An important question is whether the two unique problems posed by wireless links can be successfully addressed within the standard Internet architecture, as opposed to requiring new transport protocols designed specifically for wireless links or requiring wireless links to ‘fix up’ the operation of specific end‐to‐end protocols. We provide experimental evidence that a combination of protocol‐blind link‐level local error control, which lessens the damage, and error‐sensitive link scheduling, which ensures sensible outcomes in response to link capacity loss, provides a good operating environment while adhering to traditional Internet design practices. Copyright © 2001 John Wiley & Sons, Ltd.

While packet scheduling for wired links is a maturing area, scheduling of wireless links is less mature. A fundamental difference between wired and wireless links is that wireless media can exhibit substantial rates of link errors, resulting in significant and unpredictable loss of link capacity. This capacity loss results in a special challenge for wireless schedulers. For example, a weighted fair queue (WFQ) scheduler assumes an error free link and specifies how flows should share the link capacity. However, this specification is not sufficient to determine the correct outcome when link capacity is sharply reduced, because flows that have been allocated the same weights may differ greatly in their ability to tolerate throughput loss. In this paper, we first describe the wireless scheduling challenge in terms of an effort-outcome disconnection. Next we propose a novel notion of fairness for wireless links, effort-limited fairness (ELF), which extends WFQ via dynamic weight adjustments. ELF guarantees that all flows experiencing an error rate below a per flow threshold receive their expected service, defined as a specified rate for reserved flows or a specified share of best-effort capacity for best effort flows. After motivating and defining ELF, we present a practical approximation algorithm, which we evaluate through both trace-driven simulation and measurement of a prototype wireless radio network based on the WaveLAN physical layer.

Wireless links can exhibit high error rates due to attenuation, fading, or interfering active radiation sources. To make matters worse, error rates can be highly variable due to changes in the wireless environment. Researchers and developers have explored a wide range of solutions to optimize communication in this difficult error environment, including traditional end-to-end solutions, link-layer solutions, and solutions involving layer four processing inside the network. A significant challenge is ensuring that systems with multiple layers of error control avoid compromising performance by duplication of effort. We argue and demonstrate that protocol-independent link-level local error control can achieve high communication efficiency even in a highly variable error environment, that adaptation is important to achieve this efficiency, and that inter-layer coexistence is achievable. The logical link control layer of our WaveLAN-based experimental LAN includes three error control mechanisms: local retransmission, adaptive packet shrinking, and adaptive error coding. Measurements generated on a variety of network topologies and trace-based error environments demonstrate the TCP performance improvements and good coexistence with TCP's end-to-end retransmission strategy.

There is general belief that networks based on wireless technologies have much higher error rates than those based on more traditional technologies such as optical fiber, coaxial cable, or twisted pair wiring. This difference has motivated research on new protocol suites specifically for wireless networks. While the error characteristics of wired networks have been well documented, less experimental data is available for wireless LANs.In this paper we report the results of a study characterizing the error environment provided by AT&T WaveLAN, a commercial product designed for constructing 2 Mb/s in-building wireless networks. We evaluated the effects of interfering radiation sources, and of attenuation due to distance and obstacles, on the packet loss rate and bit error rate. We found that under many conditions the error rate of this physical layer is comparable to that of wired links. We analyze the implications of our results on today's CSMA/CA based wireless LANs and on future pico-cellular shared-medium reservation-based wireless networks.

Among the many benefits of ATM networking are the potential for connections with negotiated quality-of-service (QoS) guarantees and application-specific data management at network endpoints. We describe the architecture of a PCI bus host adapter for OC-3 and OC-12 ATM, focusing on challenges in the areas of buffer management and flow control, since these are vital to realizing the bandwidth and QoS potential of ATM endpoint hosts.