Due to the low price and robustness resulting from its wide acceptance and deployment, Ethernet has become an attractive candidate for real-time control networks. However, it is difficult to build a real-time control network using the standard Ethernet because the Ethernet MAC protocol-1-persistent CSMA/CD protocol-may cause unpredictable access delay. When both real-time and nonreal-time packets are concurrently transported over an ordinary Ethernet, real-time (RT) packets from a node may experience a large delay due to 1) contention with nonRT packets in the local node where they originate and 2) collision with RT and nonRT packets from the other nodes. To resolve this problem, we design, implement, and evaluate an adaptive traffic smoother. Specifically, we design two adaptive traffic smoothers, one at the kernel level and the other at the user level. The kernel-level traffic smoother is installed between the IP layer and the Ethernet MAC layer for better performance, and the user-level traffic smoother is installed on top of the transport layer for better portability. The kernel-level traffic smoother first gives RT packets priority over nonRT packets in order to eliminate contention within the local node. Second, it smoothes nonRT traffic so as to reduce collision with RT packets from the other nodes. This traffic smoothing can dramatically decrease the packet-collision probability on the network. The traffic smoother, installed at each node, regulates the node's outgoing nonRT traffic to maintain a certain rate. In order to provide a reasonable nonRT throughput while providing probabilistic delay guarantees for RT traffic, the nonRT traffic-generation rate is allowed to adapt itself to the underlying network load condition. Our implementation of the traffic smoother requires only a minimal change in the OS kernel without any modification to the current standard of Ethernet MAC protocol or the TCP or UDP/IP stack. The traffic smoother has been implemented on the Linux OS and is shown to reduce the RT message deadline-miss ratio up to two orders of magnitude under a heavily loaded condition, while the nonRT throughput drops only by half. For better portability, we also implemented and evaluated a user-level traffic smoother on top of the transport layer. Moreover, we emulate a switch using a kernel-level traffic smoother to evaluate the performance of traffic smoother in a switched Ethernet environment that can scale to a large control network.

In order to realize real-time communication over Ethernet or fast Ethernet, one must be able to bound the medium access time within an acceptable limit. The multiple access nature of Ethernet makes it impossible to guarantee a deterministic medium access time (hence, packet-delivery deadlines) to individual stations. However, one can bound the medium access time statistically by limiting the packet-arrival rate at the medium access control (MAC) layer. While considering automated manufacturing systems as the main target application, this paper addresses the connection admission control (CAC) problem for statistically bounding the medium access time of Ethernet. Specifically, a packet is guaranteed to have a medium access time smaller than a predefined bound with a certain probability if the instantaneous packet-arrival rate is kept below a certain threshold. Through a mathematical analysis, we first derived such a threshold. In order to keep the packet-arrival rate under the given threshold, we developed and installed middleware which 1) resides between the transport layer and the Ethernet datalink layer, and 2) smooths packet streams between them. The implementation of this middleware requires only a minimal change in the OS kernel without modification to the current standard of Ethernet MAC protocol or TCP or UDP/IP stack. In order to solve the CAC problem, we derived the probability of transmitting a packet successfully upon each trial by modeling the MAC protocol, 1-persistent CSMA/CD, and the collision resolution protocol inary exponential backoff - of Ethernet. Our in-depth simulation results have shown this analytic model to provide a reasonably accurate estimate of packet-loss (or deadline-miss) ratio over fast Ethernet. Finally, we implemented the middleware on the Linux OS, experimentally demonstrating the effectiveness of our approach in providing real-time communication over Ethernet.

Unlike deterministic real-time communication in which excessive resources may be required for "absolute" performance guarantees, statistical real-time communication seeks to achieve both probabilistic performance guarantees and efficient resource sharing. This paper presents a framework for statistical real-time communication in ATM networks, providing delay-guaranteed transport of MPEG-coded video traffic with a statistically-guaranteed cell-loss ratio. Delay-guaranteed communication is achieved with a modified version of Traffic-Controlled Rate-Monotonic Priority Scheduling (TCRM). A set of statistical real-time channels that share similar traffic characteristics are multiplexed into a common macrochannel. Those statistical real-time channels which are multiplexed together share the resources of a macrochannel, and individual statistical real-time channels are given timeliness and probabilistic cell-loss guarantees. A macrochannel is serviced by the modified TCRM which improves link utilization and makes channel management simpler. Based on the analysis of an M/D/1/N queueing system, we propose a procedure for determining the transmission capacity of a macrochannel necessary to statistically guarantee a cell-loss ratio bound. Our extensive trace-driven simulation has shown the superiority of the proposed framework to the other approaches. The overall cell-loss ratios for multihop statistical real-time channels are shown to be smaller than the predetermined bounds, thus verifying our analytical results.

To guarantee the delivery of real-time messages before their deadline, a real-time channel or connection must be established before the transmission of any message belonging to the connection. During this channel establishment phase, one must first select a route between the source and destination of this channel and then reserve sufficient resources along this route so that the worst-case end-to-end delay over the selected route may not exceed the user-specified delay bound. We propose an efficient distributed route selection scheme that is guaranteed to find a "qualified" route, if any, satisfying the performance requirement of the requested channel without compromising any of the existing guarantees. The proposed scheme can also eliminate the common reliability/performance bottleneck of a centralized route selection scheme while improving efficiency over the centralized and other distributed schemes. Although the proposed solution starts with searching all possible routes in parallel, it prunes infeasible routes quickly, and its worst-case operational overhead is shown to be only a linear function of the number of links in the network. Several examples and simulation results are presented to demonstrate the effectiveness of the proposed distributed route selection scheme as compared to sequential route-search schemes.

Unlike deterministic real time communication in which excessive resources may be required for absolute performance guarantees, statistical real time communication seeks to achieve both probabilistic performance guarantees and efficient resource sharing. We propose a framework for statistical real time communication in ATM networks that provides delay guaranteed transport of MPEG coded video traffic with a statistically guaranteed cell loss ratio. In order to provide delay guaranteed communication service, we employ a modified version of Traffic-Controlled Rate-Monotonic Priority Scheduling (TCRM). We multiplex a set of statistical real time channels that share: (i) similar traffic characteristics into a common channel called a macro channel, and (ii) the resources of the macro channel. Individual statistical real time channels are given timeliness and probabilistic cell loss guarantees. A macro channel is serviced by the modified TCRM which improves link utilization and makes channel management simpler. Using the analysis of an M/D/1/N queueing system, we propose a procedure for determining the transmission capacity of a macro channel needed to statistically guarantee a cell loss ratio bound. Simulation results have shown our framework to work well as compared to the other approaches. The overall cell loss ratios for multi hop statistical real time channels are shown to be smaller than the pre-determined bounds.

B-ISDN is expected to carry various types of traffic for which stringent quality of service (QoS) requirements must be met. In particular, per-packet (per-cell) delay guarantees are an important QoS requirement for real-time applications. Although several methods have been proposed to provide delay guarantees in packet-switching networks, they are either too complex for ATM networks which allow only very simple operations to achieve high bandwidths (hence, high-speed switching and routing), or too inefficient (in using network resources) to be cost effective. In this paper, we propose a cell-multiplexing scheme for the real-time communication service in ATM networks. The proposed scheme achieves an efficiency close to that of packet-by-packet generalized processor sharing, and incurs an implementation cost similar to that of rate controlled static priority queueing. It employs the leaky bucket model as the input traffic description model and regulates the input traffic at the user network interface to follow the input traffic specification. Inside the network, it consists of two components, traffic controller and scheduler. The function of the traffic controller is to shape real-time traffic to have the same input pattern at every switch along the path. The end-to-end delay is bounded by the scheduler which employs a nonpreemptive rate-monotonic priority scheduling policy at each switch on the path. The proposed scheme is compared to three other popular schemes with MPEG-coded movie clips. Finally, we present a hardware implementation of the proposed scheme based on a systolic array priority queue.

Future (broadband) integrated services networks are expected to carry various types of traffic including real-time packet video for which stringent quality of service (QoS) requirements must be met. We propose a cell multiplexing scheme for providing real-time communication services in point-to-point ATM networks. Our scheme is simple enough to be used in high-speed ATM networks, yet reasonably efficient in terms of accommodating channel requests. The scheme employs the leaky bucket model as the input traffic description model and regulates the input traffic at user network interface (UNI) to comply with the input traffic specification. Inside the network ATM switches provide bounded local delays to individual cells using a traffic controller and a non-preemptive rate-monotonic priority scheduler.