This paper presents the design of the automated resource management tool (ARMT), a software-based network management tool that automates the potentially complex manual task of global resource management in a mission-oriented, intranet environment. It allows the network operator to specify high level policies defining the network quality of service (QoS) that must be offered to the different applications sharing the IP network, under each operating mode of the network. The tool automatically handles resource conflicts in accordance with the high-level QoS policies defined by the network operator. It also monitors the network to detect events that may lead to deviations from the specified service policy and automatically implements proactive actions to ensure that the network delivers the required QoS. ARMT implements an innovative policy-driven QoS control technique which ensures that multimedia intranet applications will have an adaptive environment that permits them to coexist with other mission applications. The paper presents our experience with the implementation of a rapid prototype of the ARMT in a laboratory testbed environment.

Conventional routing protocols used in wireline and wireless networks are not designed to operate in an autonomous (or ad hoc) wireless local area network (AWLAN) environment with asymmetric (i.e. unidirectional) links. This paper presents the design of a routing protocol called source-initiated adaptive routing algorithm (SARA), for AWLAN environments that may contain asymmetric links. To allow efficient operation in a symmetric network while ensuring correct operation in asymmetric networks, SARA supports two different modes of protocol operation: one optimized for symmetric networks and the other providing additional capabilities needed for operation in asymmetric networks at a higher cost. The ability of SARA to adapt its mode of operation and optimize its execution for each network environment is one of its major advantages and distinguishes it from existing routing protocols for, wireless networks. Furthermore, within each of its operating modes, SARA supports best-effort as well as policy-based routing.

Network shared memory (NSM) is a low-cost approach for clustering the processing elements of commodity workstations into a single, shared-memory, mid-range parallel computer. This approach is based upon a new way of looking at the role of communication networks in multi-computer systems. The simulation results indicate that the NSM approach has the potential to outperform existing approaches for workstation clustering. In the NSM approach, each workstation has an add-on interface card, called the Network Memory Interface (NMI) card, that is used to interconnect workstations processing elements into a parallel computer. Collectively, the NMIs implement network shared memory, i.e., the coherent, shared address space of all the processors in the cluster. The authors focus on the protocol that is used by the NMIs to communicate with one another.

The authors present the design of an adaptive service for streaming MPEG video over a best-effort IP network environment. The adaptive video streaming service exploits the inherent adaptiveness of video applications to perform controlled and graceful adjustments to the perceptual quality of the displayed MPEG video stream in response to fluctuations in the QoS delivered by the three primary components in the end-to-end path of a video stream, i.e., the video server, the network, and the receiver. The design supports multiple video adaptation techniques that can be applied individually or in combination to adapt the transmitted video stream in response to fluctuations in the QoS provided by the underlying system. A novel aspect of the design is a video adaptation algorithm that selects the adaptation delivering the best perceptual quality for the video playback for a given QoS delivered by the underlying infrastructure. They describe a prototype implementation of the adaptive video streaming service and present the results of a performance evaluation of this prototype system.

In this paper, we describe a new approach for clustering workstations into a dedicated, medium-sized, shared memory parallel processor. This new approach, called the network shared memory (NSM) approach, is based upon a new way of looking at the role of communication networks multi-computer system. We develop an implementation model of the architecture of an NSM based workstation cluster. This model serves as the basis for simulations that we use to assess the performance of NSM based workstation clusters. We also use simulations to evaluate the performance of architectures representative of existing approaches for workstation clustering as well architectures representative of commercial symmetric multi-processors. The results of the performance assessment show that the NSM approach outperforms existing approaches for clustering workstations.

In this paper, we define a new approach for implementing the media transport service for video retrieval applications. We present an end-to-end transport protocol called the Video Retrieval Protocol (VRP) for transporting compressed, stored video over a wide-area ATM network to a remotely connected equipment for real-time playback of the transported video stream. VRP is designed to operate over the ABR and VBR+ services of ATM and is targeted for use by next-generation video retrieval applications. The design of VRP is based on the application of three new techniques called spectral filtering, spectral data segmentation and spectral packetizing. We used simulations to assess the performance of VRP as well as that of two existing transport protocols: TCP and UDP. The simulations were driven by a synthetic video stream. The yardstick used to compare the performance of the three protocols was their ability to preserve the perceptual quality of the transported video stream under various network congestion conditions. We show that VRP performs better than TCP and UDP in terms of the overall perceptual quality of the displayed video stream.

A discussion is presented of the issues surrounding network management functions for a proposed high-performance, highly fault-tolerant, distributed, real-time system architecture. Although many of the traditional concepts of network management such as fault, configuration, performance, accounting, security, planning, and application management certainly apply to such systems, there are additional considerations. As discussed these include technical issues such as common-mode failures and the network manager's inherent reliability, as well as development of testing and evaluation criteria for the network manager itself. Finally, modeling, simulation, and monitoring are important aspects to network manager assessment.>

This paper will take a systems-oriented view of computer communication systems. It will begin with the examination of the concept and relationships of distributed systems and networks. There are many and varied definitions of distributed systems. For purposes of this discussion we will use a simple definition with which a large number of researchers(1) tend to agree. A distributed system is defined to be a system in which a processing load is distributed across a number of processing elements (PES) which are physically interconnected, but allow varying amounts of software intercommunication. Research in the area of distributed processors is proceeding mainly in three areas: network systems(2), special-purpose systerns(3), and high-performance, general-purpose systems(4).

The purpose of this paper is to discuss the concept of hardware/software trade-offs from the perspective of the hardware designer. Primarily, the paper will examine software oriented functions/goals/requirements and discuss considerations in determining the kind of hardware support which would be involved in implementing the particular concept. This orientation follows a long-standing tradition; i.e., traditionally what appears in hardware is a result of functions which migrated into software packages, as the overall user community recognized their usefulness, and eventually migrating into hardware/firmware as manufacturers realized the usefulness of the functions. There are three dimensions to usefulness; i.e., is the function well understood, broadly used, and is the function cost effective to implement in hardware. Examples of functions which migrate from software into hardware are: floating point arithmetic, indexing, and memory management systems (virtual memory). A classic example of a function (well understood and broadly used) which has not migrated into hardware is payroll accounting. Probably the most important point is that payroll accounting routines are very specialized and many different versions (some observers place the number around 50,000) exist so that there is little standardization. This, however, does not preclude the placing of payroll accounting or other complex functions into hardware. For example, a number of researchers are currently seriously pursuing the problem of providing hardware support for data base management systems.

The real-time command and control environments characteristic of tactical military systems and industrial process control systems place unique and conflicting design requirements on a support computer system. These requirements include fast context switching, selective protection of programs and their files, controlled sharing of program and files, high processing speed, flexible, yet fast priority structure for interrupts and program execution, flexible high-speed I/O and flexible intercommunication and control for interacting programs. This paper describes a new computer system now being prototyped to satisfy these requirements. The system design includes a number of unique features relating to the efficient implementation of a flexible virtual memory addressing system for the stringent real-time environment.

This paper discusses the design of a real-time computer. The computer's design requirements, design decisions, and architecture are summarized. The paper discusses how the design requirements influenced the computer architecture. The system's three upward compatible addressing options (real, base, virtual) are also discussed.

This correspondence considers the cost and performance factors in computer architecture and how they apply to parallel processors.

A system design is given for a computer capable of efficient execution of APL source statements and programs. The system is organized in a new type computer architecture based upon the storage of matrices and vectors in cellular logic arrays. Almost all of the scalar, mixed, and composite fixed point APL operations are implemented on the machine by algorithms given in this paper. The detailed logical design of the machine has not been completed; however, all the needed logic and memory has been specified. Also, the functions that each hardware cell must be able to perform have been specified. The algorithms for most APL operations are presented in flow charts or tables which give the detailed data flow produced by the execution of each APL operation in the machine.