Showing papers on "Internetwork protocol published in 1991"

B-ISDN architecture and protocol

Abstract: The authors discuss the architecture and protocol for broadband ISDN (B-ISDN) based on the CCITT standards. The discussion attempts to address the general concept of B-ISDN architecture and protocol and, whenever possible, present alternatives and the rationale for decisions in the selection of the protocol. B-ISDN is presented as a network evolution, and the impact of the asynchronous transfer mode (ATM) on the network is described. The role of virtual channel and virtual path in B-ISDN is discussed. The B-ISDN protocol structure and lower layer functions comprising the physical, ATM, and adaptation layers are presented, and the tentative trends of signaling and traffic control for the B-ISDN are delineated. >

Encryption with selective disclosure of protocol identifiers

Abstract: A method for selective disclosure of the identity of a communication protocol under which an information packet originated, but without incorrectly identifying the protocol in a header accompanying the packet. If there is a need to conceal the identity of the underlying source protocol, a special anonymous protocol identifier is used, instead of the real protocol identifier, in the header of an encrypted information packet. Network monitors can then still provide accurate information concerning traffic on the network, without having this information distorted by the use of incorrect communication protocols. If there is a desire to reveal the underlying protocol, a subnetwork protocol frame format is used to store the protocol identity and signify whether the packet is encrypted. A packet that is of a non-subnetwork protocol can be encapsulated with a subnetwork header containing a special code signifying that there is an encapsulated packet and containing the original protocol identifier.

Relationship of the Signaling System No.7 protocol architecture to the OSI reference model

Abstract: The relationships that currently exist between the Signaling System No.7 (SS7) and Open Systems Interconnection (OSI) architectures are examined. The functions of each layer in the SS7 protocol stack, as well as the SS7 addressing mechanisms, are described to the extent necessary to show correspondences to those of OSI. Those areas where the two architectures do not align are highlighted. The criteria by which alignment of SS7 with OSI standards is judged are explained. Work in progress in CCITT in defining the evolution of various SS7 application layer standards, where a conscious effort is being made to attempt to align with OSI standards, is described. >

An approach to indirect protocol conversion

Abstract: With the proliferation of different computer networks, protocol conversion is needed to achieve interoperability between computer networks that implement different protocols, so that users on different networks can communicate with one another. In its most efficient form, protocol conversion is performed on the fly, i.e. the messages produced in one protocol are converted and passed immediately to the other protocol; this is called direct conversion, and no buffer is needed to store the messages. For many practical protocols, however, it is possible that the messages produced in one protocol need to be re-ordered and converted before they are passed to the other protocol; this is called indirect conversion, and the messages produced in one protocol must be first stored in a non-FIFO buffer and then converted in proper order for the other protocol. This paper presents a formal technique to synthesize a converter by inserting the PUT and GET operations into two given protocols to manipulate the buffer for indirect conversion. The PUT operation inserts messages in one protocol into a non-FIFO buffer, and then the GET operation retrieves the messages in proper order from the buffer and converts them for the other protocol. Since the PUT and GET operations are asynchronous, they may create some problems for the buffer: (1) buffer overflow, i.e., the number of messages stored in the buffer exceeds its capacity; and (2) improper termination, i.e., the protocol terminates with some messages leftover in the buffer. Therefore, the synthesized converter needs to be validated. Several techniques are proposed to reduce the computing time and space spent in validating the converter. Finally, an algorithm for indirect conversion is presented and applied to a practical example for conversion between ISO OSI and IBM SNA protocols.

Realistic simulation of a broadcast protocol for an inter vehicle communication system (IVCS)

Abstract: A novel multicast protocol of a short-range radio communication system for applications in civil road traffic is introduced. General communication characteristics for a road traffic radio LAN (local-area network) are described. The protocol algorithm is based on the periodic transmission of data and the indication of missing data performed by NAKs (negative acknowledgements). The protocol algorithm is specified, and analytical and simulation results are presented. It is shown that the cost of such a protocol in terms of additional required bandwidth is remarkably low. >

A parallel implementation of the ISO 8802-2.2 protocol

Abstract: A parallel implementation of the ISO 8802-2.2 Logical Link Control protocol on a multiprocessor-based communication adapter is described. Detailed measurements allow the authors to construct and parameterize a model of the implementation. The model shows how pipelining of different protocol functions is possible and what performance gain can be expected when running the implementation on different processor configurations. The performance of the parallel implementation is more than 16000 information protocol data units per second, commensurate with emerging high-speed networks operating in the 100 Mb/s range. >

Reliable Interconnection Networks for Parallel Computers

Abstract: This technical report describes a new protocol, the Unique Token Protocol, for reliable message communication. This protocol eliminates the need for end-to-end acknowledgments and minimizes the communication effort when no dynamic errors occur. Various properties of end-to-end protocols are presented. The unique token protocol solves the associated problems. It eliminates source buffering by maintaining in the network at least two copies of a message. A token is used to decide if a message was delivered to the destination exactly once. This technical report also presents a possible implementation of the protocol in a worm-hole routed, 3-D mesh network.

An Expanded Spanning-tree Protocol For Home-oriented Network Management

Abstract: An expanded spanning-tree protocol capable to simultaneous configurations of plural logical spanning-trees is proposed. It configures a set of optimum spanning-trees and offers an optimal routing path to each packet. This protocol should be highly effective for minimizing the possible traffic increase caused by routing packets and for minimizing the transmission lag. Details of the operating algorithm of the bridge are described. Also proposed is a simplified version of this protocol, called the approximate protocol, which can be incorporated in a more compact form and is much easier to apply. The benefits of these proposed protocols are quantitatively evaluated in terms of the reduction of traffic and packet delay compared with the conventional protocol. >

Communication system having ISDN signalling capability

Abstract: A communication system operates compatibly with a public or private switching network using a Q.931 protocol for basic rate ISDN communications. Communications between a control unit and station terminals of the system use a local protocol which is a modification of the Q.931 protocol. This local protocol includes one segment including the mandatory elements of the Q.931 protocol and a second segment derived by converting the optional elements of the Q.931 protocol into a local codeset which controls the user interface circuits at each station terminal. The local protocol enables call control between the control unit and each station terminal to be managed on a bearer channel basis while call control between the system and the network is managed on a call appearance basis using the standard Q.931 protocol. The local protocol provides for communicating line selection and feature activation commands from the terminal to the control unit, using simple button commands.

Connection management in a LAN to broadband ISDN gateway: a simulation study

Abstract: An examination is made of a gateway, BERGATE, designed and developed as part of the broadband ISDN pilot BERKOM project of the German PTT. End systems support the ISO 8073 transport protocol on top of the ISO 8473 internetwork protocol. Each BERGATE may access a single circuit switched H4 channel. Data exchange between a pair of gateways is possible only after establishing an H4 connection between these gateways. Since ISO 8473 is a connectionless protocol, it gives no indication when to open and close the H4 connection. The performance of a simple connection management policy is investigated, and proper values of parameters for this policy are selected. >

Fairness issues for mixed TCP/OSI internets

Abstract: The author focuses on experiments and simulations designed to examine the difficulties in providing a fair communication service in an internet carrying traffic from the transport control protocol (TCP) and open systems interconnection (OSI) applications. Fairness between connections that compete for resources is a goal of congestion control schemes that have been developed for both protocol suites. The slow start congestion control mechanism is now fairly standard for TCP, and the DEC-bit approach to congestion avoidance is widely accepted for OSI networks with a connectionless network layer. Unfortunately, these two approaches have evolved independently and are not designed to provide fairness between protocol suites. The network experiments and simulations reported examined the extent of fairness problems to be expected in a mixed TCP/OSI environment and the effects of a simple router-based measure to limit unfairness between protocol suites. >

Protocol conversion for computer networks

Abstract: With the proliferation of network architectures, protocol conversion is needed to achieve interoperability between different computer network protocols. To accomplish a protocol conversion, semantic equivalence between the messages of two protocols has to be identified; and then a converter is constructed to convert the messages of one protocol into semantically equivalent messages for the other protocol. In its most efficient form, protocol conversion is performed on the fly, i.e., the messages produced in one protocol are converted and passed immediately to the other protocol; this is called direct conversion, and no buffer is needed to store the messages. For many practical protocols, however, the messages produced in one protocol need to be re-ordered and converted before they are passed to the other protocol; this is called indirect conversion, and the messages produced in one protocol must be first stored in a non-FIFO buffer and then converted in proper order for the other protocol. In this dissertation, the direct conversion problem is resolved by inserting coupled points into two given protocols. When a coupled point is reached in one protocol, it has to wait until the other protocol reaches the same coupled point; then a pair of OUT and IN operations, constituting the coupled point, convert and pass a message from one protocol to the other protocol. In contrast to the synchronous communication between OUT and IN operations for direct conversion, asynchronous communication operations PUT and GET are inserted into two given protocols to manipulate the buffer for indirect conversion. The PUT operation inserts messages in one protocol into a non-FIFO buffer, and the GET operation then retrieves the messages in proper order from the buffer and converts them for the other protocol. However, the PUT and GET operations may create some problems for the buffer: (1) buffer overflow, i.e., the number of messages stored in the buffer exceeds its capacity; and (2) improper termination, i.e., the protocol terminates with some messages leftover in the buffer. Therefore, the synthesized converter needs to be verified. Several techniques are proposed to reduce the computing time and space spent in verifying the converter.

Performance of connection management policies for a LAN to broadband ISDN gateway

Abstract: We consider the problem of connecting LANs via ISDN links which are connection-oriented in nature. Data exchange between a pair of gateways is possible only after establishing an ISDN connection between these gateways. It is assumed that LAN endsystems support a connection oriented transport protocol on top of the connectionless Internetwork Protocol which does not provide the gateway with any indications when to open and close the ISDN connection. Thus there appears a need for some specific policy for ISDN connection management. We present a systematic investigation of the connection management strategies for a case study: a gateway BERGATE designed and developed within a pilot broadband ISDN implementation BERKOM supported by the German PTT. Assuming that the ISDN interface allows only a single connection, we investigate the performance of a simple connection management policy, select proper values of parameters for this policy, and demonstrate that this policy performs only marginally worse than an ideal policy using full information about the state of transport connections.

Implementation of OSI-IP router for LAN-LAN interconnection

Abstract: An implementation of a router with a capability to dynamically upgrade its routing table based on ISO 8473 and ISO 9542 standards is described. It is suited for use in OSI CLNS networks similar to US GOSIP recommendations. It is being proposed to use this router as a parallel stack in an existing TCP/IP router so that both OSI and TCP/IP network services can be supported over a campus wide network.

A performance comparison of connection-oriented and connectionless LLC protocols in a high-speed satellite data network

Abstract: A performance comparison is made between a connection-oriented (CO) logical link control (LLC) protocol and a connectionless (CL) LLC protocol in a high-speed satellite packet network with the TDMA-reservation multiple access protocol. In particular, the authors examine how the channel transmission rate affects the performance of the two LLC protocols and then determine which LLC protocol is more suitable for any given system environment. The comparison indicates that the CL-LLC protocol has more advantages over the CO-LLC protocol as the channel transmission rate increases and that the suitable region for the CL-LLC protocol becomes wider as the receive-buffer capacity increases and as the network load decreases. >

IAPA: a new protocol architecture based on interacting activities

Abstract: As the throughput of communication channels grows to the order of GBits/s, communication protocols become the bottleneck in communication applications. The reason for this bottleneck resides partly in the layered protocol architecture, and partly in the complexity of the protocols. This paper presents a new protocol architecture that potentially reduces the protocol overhead by providing a more flexible design structure, giving the opportunity to design protocols suitable for efficient, parallel implementation. >

Integrated routing for multi-protocol TCP/IP-OSI environments

Abstract: This paper describes an Integrated Routing Protocol, based on the OSI IS-IS protocol, which may be used as an Interior Gateway Protocol (IGP) to support TCP/IP and well as OSI CLNP. This allows a single routing protocol to be used in pure IP environments, pure OSI CLNP environments, as well as multi-protocol environments.

Specification and verification of real-time broadcast networks

Abstract: A real time communications protocol satisfies certain timing constraints in order to meet its service specification. A new generation of real-time protocols has been proposed for broadcast bus networks that operate at very high speeds (50-200 Mbps). These protocols exploit the directionality of signal propagation and impose stringent timing constraints to provide collision-free access to a shared bus within a bounded delay. These features make these protocols particularly suitable for real-time applications, such as the transport of integrated data, voice, video and facsimile traffic. The behavior of a typical real-time protocol is complex and, to have reasonable confidence in its implementation, it has to be analyzed using a formal method. Although a number of real-time protocols for broadcast bus networks have been proposed, there is no formal method for their analyses. We present a model for specifying and verifying real-time protocols for high-speed broadcast bus networks. In this model, the network topology and broadcast bus behavior is specified by a set of channel axioms that model the actual propagation of signals with time along the bus. Protocol entities are specified as sequential programs in a Pascal-like specification language that includes two novel wait constructs. The interaction between channels and protocol entities is modeled by shared variables. Real time is modeled by a global clock, time variables and a time event. A first-order logic, that includes time variables, is used to reason about the temporal behavior of the protocol. To illustrate our model and verification method, we specify two protocols proposed in the literature, the EHS protocol and the Expressnet protocol. The topology of the EHS protocol consists of a bidirectional broadcast bus and a unidirectional control wire. We prove that the protocol is collision-free using an assertional proof method. The topology of the Expressnet protocol consists of a folded unidirectional bus. We apply a proof method based on timed-reachability analysis to the Expressnet protocol. Our analysis shows that the original Expressnet protocol is not collision-free. To achieve collision-freedom, we propose a modification to the protocol. We show that the modified protocol is collision-free and also derive a bound for the access delay.

An analysis of OSI in the NASA Science Internet

Abstract: An analysis is presented of issues surrounding the integration of the open systems interconnection (OSI) protocols into an operational national network. The OSI protocols are evolving as international standards and they have been adopted by the US government as a Federal Information Processing Standard (FIPS). Yet, a complete OSI protocol stack is still under development, profiles for options within the protocol suite are yet to be defined for government networks, and product availability is somewhat sparse. In addition, there are issues to address when wide area networks that are disjoint due to the use of different networking protocols all start using common protocols, particularly in the area of routing. These items are studied from the perspective of a large operational network (the NASA Science Internet). >

Communication protocol for a local network based on code division multiplexing

Abstract: A communication protocol for a local network that handles voice signals and uses ‘direct sequence spread spectrum’ techniques ( DS-SS), is presented. The control of the protocol is distributed among the network users, providing them with the capability for full-duplex communication and successful channel acquisition without data collisions. The access of the single communication channel by any user of the network and the routing of calls are based on the detection of the number of the decoded signal levels. The implementation of the network is simplified with the use of a single code per communication. The protocol operation ( including the call scheduling and signalling processes) that generates all the signals necessary for communication in the network ( ring, idle, busy) is described in detail.

Simulation and evaluation of FTP and TCP/IP on FDDI backbone network

Abstract: Presents an evaluation of the TCP/IP (Transmission Control Protocol/Internet Protocol) under various traffic conditions by simulating file transfer between two Ethernet stations connected via an FDDI (Fiber Distributed Data Interface) backbone. The model is built using a simulation language called Research Queueing Package. The results presented indicate that TCP works slower for smaller files, partly because of the chosen parameter values (e.g. buffer sizes, the initial window size, the initial timeout value) and partly because of the long wait for the retransmission timeout to expire. The simulation results and data from a real FTP (file transfer protocol) show that TCP/IP is sensitive to different parameter values, especially the timeout values. It is also found that TCP introduces a significant overhead. Therefore, it is concluded that TCP is not very efficient for strictly local area networks, and that a more efficient protocol for this purpose must be developed. >