In this paper the ETMUCS project with its novel network interface is introduced. The requirements of the communication system are elaborated, based on the functionality described. The implementation of the Interaction...
详细信息
In this paper the ETMUCS project with its novel network interface is introduced. The requirements of the communication system are elaborated, based on the functionality described. The implementation of the Interaction- Activity Model in hardware and software is briefly presented. The Network Interface Module is the core of the network interface. It bears some unusual features that can be exploited by the current node (switch) of the high speed fiber optic network.
In this paper we describe design considerations for the implementation of OSI protocols on parallel processors. We consider parallelism to be an appropriate technology for implementing protocols for high-speed network...
详细信息
In this paper we describe design considerations for the implementation of OSI protocols on parallel processors. We consider parallelism to be an appropriate technology for implementing protocols for high-speed networks. Parallelism can help to overcome the gap between fast fiber optics and slow protocol execution.
This paper describes our experience in implementing the OSI transport service, with the focus being on the use of the PASS ("Parallel Activity Specification Scheme") Formal Description Technique (FDT) tools ...
详细信息
This paper describes our experience in implementing the OSI transport service, with the focus being on the use of the PASS ("Parallel Activity Specification Scheme") Formal Description Technique (FDT) tools for specification, verification, and semi-automatic code generation. We discuss the results of the experiment from two aspects: (1) efficiency of implementation, and (2) efficiency of the resultant implementation. The PASS tools, recently augmented with a graphical user interface, are described, and their application to the present implementation is outlined.
The brief paper outlines our strategies for providing a hardware and software solution to interfacing hosts to high-performance networks. Our prototype implementation connects an IBM RS/6000 to a SONET-based ATM netwo...
详细信息
The brief paper outlines our strategies for providing a hardware and software solution to interfacing hosts to high-performance networks. Our prototype implementation connects an IBM RS/6000 to a SONET-based ATM network carrying data at the OC-3c rate of 155 Mbps. We have measured application-to-network data rates of up to 130 Mbps.
This report addresses issues we faxed in our on going work at the Radiology Research Laboratory. These issues are related to communication protocols for very high speed local area networks in a medical imaging environ...
详细信息
This report addresses issues we faxed in our on going work at the Radiology Research Laboratory. These issues are related to communication protocols for very high speed local area networks in a medical imaging environment. In this report we describe the characteristic of the application, solutions utilized in coupling between host system and the network, and protocol design.
Nectar is a multicomputer built around a high-bandwidth crosspoint network. Nodes are connected to Nectar using network coprocessors (CABs) that do the protocol processing. The communication coprocessor is based on a ...
详细信息
Nectar is a multicomputer built around a high-bandwidth crosspoint network. Nodes are connected to Nectar using network coprocessors (CABs) that do the protocol processing. The communication coprocessor is based on a general-purpose processor and runs a flexible runtime system. This allows us to offload application on the coprocessor and to experiment with different protocol implementations. An important network parameter for applications is the latency for short messages. In this paper we present an analysis of the CAB-CAB and host-host communication latency over Nectar for a number of communication protocols. We break up the latency and discuss how the components an influenced by the protocol implementation and node architecture.
The problem of processing packets in a communication node occurred due to the introduction of fiber optics as transmission medium, because the data rates increased drastically but the processing power did not increase...
详细信息
The problem of processing packets in a communication node occurred due to the introduction of fiber optics as transmission medium, because the data rates increased drastically but the processing power did not increase in the same order of magnitude. For that reason, it is important to defuse the processing bottleneck of communication controllers and transport systems in general. Different approaches have been pursued in the research of high speed communications. This paper presents the architecture of an efficient transport system supported by a developed concept called Protocol Configuration Manager (PROCOM). This mechanism allows the dynamical adaptation of the running protocol stack on explicitly requested application requirements.
The demand for highperformancecommunicationsubsystems has rapidly increased during the last few years. Emerging applications (e.g., multi-media applications) requiring highperformance are one of the driving factor...
详细信息
The demand for highperformancecommunicationsubsystems has rapidly increased during the last few years. Emerging applications (e.g., multi-media applications) requiring highperformance are one of the driving factors for this development. But still there is a performance gap of at least one order of magnitude between the bandwidth available on the network and the application requirements. This is based on the insufficient processing power inside communication nodes. They cannot serve networks with data rates in the range of Mbps to Gbps. To overcome this so-called transport bottleneck highperformancearchitectures for communicationsubsystems have to be designed considering hardware and software structure as well as protocol design and service requirements. This paper presents an approach towards highperformancecommunication platforms based on parallel protocol implementations on transputer networks. Thereby standard OSI protocols (TP4 and CLNP) as well as emerging lightweight protocols, such as XTP have been explored. The TP4 implementation is here presented more in detail and some experiences gained within this project are summarized.
The Macintosh Telemetry and Command Workstation (MacTAC) was developed by Clemson University in cooperation with NASA's Gadded Space Flight Center to meet the needs of scientists requiring low-cost, high performan...
详细信息
The Macintosh Telemetry and Command Workstation (MacTAC) was developed by Clemson University in cooperation with NASA's Gadded Space Flight Center to meet the needs of scientists requiring low-cost, highperformance telemetry workstations. Among its many functions, the MacTAC system employs Reed Solomon(255,223) error correction as recommended by the Consultative Committee for Space Data Systems (CCSDS) [1] in a pipelined hardware platform. Presented here is an improved Reed-Solomon (RS) architecture to provide real time correction of CCSDS frames at a sustained rate of 5 megabits per second (Mbps).
暂无评论