Search

Russian researchers have developed SATM ( synchronous asynchronous transfer mode), a new telecommunications protocol that supports data, voice and video transmissions more effectively than current technologies.

A mechanism for resource reservation in advance (ReRA) was created for use in distributed multimedia applications. The ReRA mechanism works by monitoring and reserving, if possible, the available resources at all involved intermediate components and end systems. Implementation of the mechanism has proven its benefits in resource management and quality of service.

Port-based testing and group-based testing are two synchronizable testing strategies that can be used for a finite state machine (FSM). Port-based testing, which has lesser limitations than pair-synchronizable testing, does not permit FSM testers to communicate with each other. On the other hand, group-based testing separates FSM ports into mutually disjoint group and permits ports testers to communicate with each other.

Two synthesis methods were formulated that can be used in designing discrete event systems (DES). The first method takes into consideration temporal limitations associated with DES design. Its effectiveness has been proven in the design of X.25 protocol extension. The second method, which utilizes concepts of controllability and coobservability, has been proven effective in solving problems associated with telephone network services.

A mathematical evaluation model was designed to assess nonsymmetric token ring network supporting stations with single packet buffers under a timed-token protocol, such as the Fiber Distributed Data Interface system. The queueing behavior of the network was assessed by characterizing sets of discrete-time Markov chains at the instants of token arrivals at the stations.

Modifying the Media Access Control (MAC) protocol for dual bus high-speed networks can result in a waste-free congestion control. In principle, such a system would allow heavily loaded nodes to be given faster transmission rates over uncongested nodes, with the uncongested nodes transferring their packets to their buffers until the time that congestion is eased. Coupling the system with a priority mechanism would enable the system to work within a multi-priority environment.

An analysis of the throughput behavior of stream protocols in admission controlled bearer networks shows that traffic shaping, a system wherein information units are delayed from entering the network by being placed in a buffer where its priority is scrutinized according to rate control parameters, reduces the chance of degradation. Transmission control protocols can be seriously affected by traffic characteristics or network response which leads to some form of degradation.

An algorithmic approach was adopted for the construction of protocol converters. The method provides a formal message mapping set model for representing the conversion requirements for the converter. The model is easy to build and permits the specification of different message mapping forms. The approach ensures the generation of a converter that does not cause unspecific receptions or deadlocks if a converter that fulfills the conversion requirements exists.

A standardized approach and a software prototype make up a generic framework for interfacing HyperText Transfer Protocols to Relational Database Management Systems (RDBMS). Such links are usually in demand because of the World Wide Web service's marked expansion. The approach is primarily based on the Query Specification File (QSF), whose design philosophy was shown via a short description of its sections. The retrieval of information from RDBMSs is addressed.

A study details the Semantic Web Graph Layer (SWGL) and its architecture. The SWGL permits the dynamic interpretation of link and node structures according to graph models. The graph model for Colored Petri Nets was implemented to demonstrate the usefulness of SWGL. The concept of browsing streams is extended to allow synchronization beyond those of a single user, enabling authors to write documents that continuously synchronize the browsing paths of multiple users.

The scope of a locally maintained hyperlink database studied in a previous research effort on an integrated scheme for an intelligent publishing environment is extended to include the entire World Wide Web. The addition of hypertext databases to the Web infrastructure has a positive effect on link maintenance, referential integrity, navigation and visualization. An architecture and migration path was developed for the deployment of the Atlas scalable hyperlink database.

LispWeb is a specialized HTTP server that delivers Distributed Artificial Intelligence over the World Wide Web. LispWeb implements the standard HTTP protocol and offers a library of high-level Lisp functions to dynamically generate HTML pages. It also offers a facility for creating Graphical User Interfaces on the WWW through dynamically generated image maps and a method for server communications. LispWeb was designed to function as the front-end of an intelligent agent network where communication is done via an HTTP extension protocol.

The go-back-n automatic repeat request protocol (ARQ) or error recovery and the sliding window flow control protocol manifests quasi-independence properties under certain conditions. These include situations characterized by independent packet errors and deterministic or exponentially distributed transmission times and delays. The characteristic, however, does not hold for the selective repeat ARQ protocol.

A hop-by-hop predictive protocol for congestion control of high-speed, real-time, multimedia wide area networks is introduced. Based on the exchange of information between adjacent network nodes, it exploits the store and forward properties of the network to forecast the 'buffer content of a node one round trip into the future.' Due to these, it is superior to static or adaptive end-to-end rate control policies.

The Hybrid Random Access and Reservation Protocol (HARP) for wavelength division multiplexing (WDM) networks configured in a star architecture was developed. HARP, which provides a high throughput-delay performance independent of the load conditions, may be implemented in WDM passive star networks using the ALOHA protocol. This, however, requires the installation of an array of polarization independent acoustically tunable optical filters at the network hub.

An assessment of the IP multicast architecture's performance showed the need to pay extra time for multicast packet processing. However, network layer multicast is still better than the emulation of multicast by point-to-point messages in terms of efficiency. The OPNET network simulation tool was used to model the IP multicast architecture. Pruned multicast was shown to possess advantages over truncated broadcast delivery.

The overhead of the HTTP protocol used in gaining access to the World Wide Web (WWW) can be greatly reduced by quantifying long-lived TCP connections and using the GETALL method. These modifications will allow a modified WWW client to interoperate with an unmodified server. They will also reduce the number of process creations done by the WWW's multiple-process servers.

A non-FIFO (first in first out) routing may implement a semi-FIFO service without large-scale modifications in the routing layer. The logical link control protocol for reliable data transmission via a semi-FIFO service uses a single logical timer and attains maximum capacity even without estimates of the round trip delay.

The sequenced synchronized clock message protocol is a timer-based connection management protocol for reliable ordered at-most-once message delivery. Multiple outstanding messages are allowed and there is no need for statically preallocated memory to store connection records. The absence of timing constraints frees its correctness from transmission rate conditions.

Two variations of the local area network structure have been developed which adhere to the general concept of the ethernet. Called the Single Channel with Segmentation and the Dual Channel with Segmentation (DCS), both variations exhibit the same properties as a standard ethernet, except that the 'ether' functions as a discontinuous medium. During simulations, the DCS exhibited a maximum asymptotic throughput which was equal to the capacity of the medium.

Most communications protocols used on multiple access broadcast networks (MABN) are unable to integrate voice and data traffic. In addition, protocols such as movable-boundary and Global Queue require significant modifications to existing data stations. To solve the voice and data integration problems on MABNs, a new protocol called the superchannel was developed. The superchannel is fully-compatible with existing MABNs and allows data stations to operate using their original protocols.

The Cyclic Queue-connection (CQ) and the CQ-class admission control algorithms for handling broadband packet network traffic under different service conditions are introduced. The two differ in that the CQ-connection segregates traffic based on individual connections while the CQ-class segregates traffic based on traffic class. A performance comparison between the two and existing algorithms is also carried out via simulations of their capabilities to handle homogeneous and heterogeneous traffic.

A computer network protocol for three different cases of sequential process transport with a timing-based priority mechanism for synchronous and real-time applications is introduced. The protocol, which groups packets according to their priority classification, was evaluated by computation of throughput and average packet delay at a node under different service conditions on a computer simulated static logical ring. It features commitment control via an immediate acknowledgement scheme.

The University of Illinois' National Center for Supercomputing Applications is working on a computer network protocol to enhance group and public annotation functionality of its HTTP server and Mosaic client. To ensure the function's scalability, a solution is proposed. It mandates each server to direct the client to sites where it can obtain public annotations for a document. In addition, it requires users to direct clients to sites where they can obtain group annotations.

New services may be added to the World-Wide Web by creating a 'layer of extended functionality' beneath the HTTP layer. Known as a DCE Toolkit Layer, its services include transport-independent communications, security, naming and other features that are supported by the OSF Distributed Computing Environment software. In addition, it provides security and location-independent hyperlinks without revising HTTP protocols.

A line of ARQ protocols is studied for possible implementation via hardware with limited state information and some FIFO buffers. The protocol is found to be similar to a normal Go-Back-N when the hardware has one FIFO buffer. A full selective repeat is achieved with multiple FIFO buffers. Performance may approximate a selective repeat via two or three FIFO buffers for common error rates. A full description and a throughput analysis of the ARQ protocol family are presented.

A new selective-repeat Automatic Repeat Request protocol called High-Speed Selective-Repeat (HSR) has been developed. HSR provides a rapid error recovery mechanism and ensures that a correctly received packet is not retransmitted. Its simplicity enables minimal computational load on the transmit and receive processes. Results showed that the protocol's performance is comparable to an ideal selective-repeat provided its parameters are adjusted properly.

FAITH is a general purpose protocol test system for integrated services digital networks (ISDN). FAITH is currently being used by PTT Telecommunications of the Netherlands for testing switches and terminals and for commercial test services for ISDN terminals. The test system supports test specification, test execution, automatic test implementation, automatic test report generation, multi parity testing methods and test case selection strategies.

The multi-token ring network protocol is proposed to compensate for the inefficiencies found in single token ring networks. The multi-token ring network is an extension of a two-token access protocol model but retains the transmitter, token manager and receiver processes. Tests show that the protocol is better than the single-token ring protocol under light to medium loads and is only slightly worse at higher loads. Results also reveal that increasing token quantity improves performance at light load but degrades performance at high load.

A study was conducted to develop schemes for improving time management in the X.25 layer 2 protocol. The aim was to reduce excessive overheads caused by improper time management implementation. A method of choosing the appropriate value of the piggyback timer was developed based on cases wherein the protocol's traffic pattern could be predicted. The method is believed to be effective in cases of heavy traffic.

Signalling System Number 7 (SS7) Network Services Part and X.25 are sometimes used interchangeably despite the differences in their stated aims. SS7 aims to provide an internationally standardized general purpose Common Channel Signalling system and is therefore designed for telephone signalling, while X.25 is designed particularly for general data communications. Caution should therefore be exercised when using one standard in place of the other.

A variation over a proposed scheduling-delay access mechanism scheme is evaluated. The modified scheme is adapted as an attempt to piggyback onto a CSMA/CD protocol. Both the original and the variant techniques are examined for average delay. Results show that average delay performance is conditioned by communications traffic. The CSMA/CD can be enhanced for time-critical applications in cases of collision.

A multimedia transport protocol (MTP) which supports real-time audio/video transmission during multimedia conferencing across packet-switched networks has been developed. The protocol uses queuing, error correction and network congestion monitoring mechanisms together with a facility for varying audio-video synchronization. It is a 'best effort' protocol which controls the negative effects of jitter, packet loss and load variation. The MTP supports computer videoconferencing through small networks using readily available audio/video technology and without specialized network services.

A connection management protocol known as Connection Management with Synchronized Clocks (CMSC) was analyzed. A connection is the association between two users that exchanges messages and the connection management protocol keeps information about the state of a connection in a connection record. CMSC was determined to be an effective protocol with several strong points including fast connection, no connection setup latency and immediate resumption of a communication after a system failure.

Two collision protocols, of the Piggyback Ethernet type, are presented for use on fast bidirectional bus networks. Good performance is found for the hybrid protocols under various load conditions. The protocols include Ethernet characteristics and also properties of schemes involving the scheduling-delay access mechanism. Protocols and networks have been modeled with the LANSF simulation package. Hardware requirements for the protocols are said to be fairly modest, compared to other protocols for fast networks.

A short overview of gigabit networking is outlined, focusing on the higher-level questions motivating gigabit research. The discussion is centered on introducing research on the subject such that results from such studies will be easier to interpret and understand.

Protocol visualization is analyzed and applied to Estelle. The Graphical Representation Of Protocols in Estelle (GROPE) is proposed as a workstation tool as a formal description technique (FDT) to identify computer communication protocols. The advantages of user manipulation of graphic complexity in Estelle specification simulation are exhibited in the use of GROPE as an interface with a standard FDT.

An analysis of global channel allocation policies in relation to policies on local sharing of channels within a local area network group (LANG) shows that these policies conform to each other. Channels within a LANG can be limited to 5 to 10 channels and still attain optimum operational performance. The Dynamic Resource Allocation for Metropolitan Areas system attains gigabit network speed and can be applied to metropolitan area networks, parallelizing the communication process for optimum network performance.

Two broadcast star network models are proposed with small and large propagation delays and a collision-avoidance switch each. The proposed service discipline integrates first-come first-served and random-order of services disciplines with reservation and priority fields. Results show that the network models can integrate random access with optimum network use and that the proposed service discipline, compared with others, is easily implemented and has low message delay variance.

An existing voice and data integration protocol for a unidirectional fiber optic network is simplified through decreased processing and storage complexity in connection with protocol parametric solutions, increased voice stations with multiple voice packets buffered at each station. Results show that the protocol maintains 100% channel use and data stations in minimum to average traffic can be accommodated equally. However, congestion occurs in heavy loads and, although multiple buffers increase voice transmission, it does not have any significant effect on overall performance.

A proposed framework is used to analyze the implementation process as part of the design process involving the successive transformation of an abstract specification into a compilable form acceptable in a computer environment. The linkages between transformations are termed implementation relations which are, in effect, formal expressions of validity. Applied to LOTOS, the method shows that implementation relations are complex, particularly in non-transitive cases. Paradoxes are revealed as well as other research aspects.

The register-insertion media-access control protocol with destination removal is recommended for advanced and future high-speed local area networks. It is shown that partitioning ring networks significantly improves maximum throughput and average response time. A new analysis method for measuring register-insertion ring network average response time is used. It provides realistic performance estimates independent of traffic load and station number.

An analysis of the unpredictable behavior of the distributed queue dual bus suggests criteria for effectivity of enhancement schemes in metropolitan area network overload conditions. The Bandwidth Balancing Mechanism adopted by the Institute of Electrical and Electronic Engineers proved to be inadequate. However, the proposed Symmetric Bandwidth Balancing scheme is believed able to meet these criteria and provide effective prioritization.

Abstract Syntax One (ASN.1) is a formal data specifications standard that complies with the Open Systems Interconnection protocol. It is used for precise information definition and formatting, a necessity for distributed network applications information interchange. Research and development of ASN.1 and its encoding rules are still ongoing. Some notable fields of research include macro definitions, object classes, constraint specifications and additional encoding rules definitions.

A study was conducted to show that a File Transfer Protocol proxy cache supporting knowledge of national mirror sites minimizes the amount of information needed to be transferred across overused networks. Biases associated with FTP traffic were analyzed using information on the use of international bandwidth and data on document access patterns. Results indicated that such proxy caches supported the transfer of data across overused networks.