This paper was published in the Proceedings of ICCC '80, pp 517-22.
K. Tomaru, T. Kato and S.I. Yamaguchi
Hitachi, Ltd., Japan
A private integrated communication network for facsimile, telex, data communication, etc. for internationally working firms was developed. A key component of the network is a packet switching node accommodating CCITT X.25 protocol. New service facilities that are performed by the use of storage facility in the node are provided. A set of protocol, placed on the upper layer of X.25 level 3, is defined to carry out these new services. In order to assure highly reliable communication, message numbering, delivery confirmation, generation and delivery of message lists, and other facilities are provided. In this paper, system structure protocol, message management procedure, etc., of the network are described.
In conjunction with the rapid progress of computer and communication technologies, electrical communication facilities in the business field have changed considerably. Several years ago, only the telephone and telex were used as main telecommunication media among a head office and branch offices of a firm. But recently, facsimile and computer communication facilities have been added to the telephone and telex.
Each type of communication is usually carried out on different circuits or networks. But from an economical point of view, it is desirable to use the same communication media such as a transmission system and a switching system, for the various kinds of communication. Particularly an international firm that has offices around the world requires a communication network in which an expensive international communication circuit can be used efficiently for various kinds of communication. In order to satisfy such a requirement, Hitachi Ltd. developed a private packet network named HIPA-NET . Usually, leased circuits are used in HIPA-NET. But interworking between HIPA-NET and a public packet network has been considered in the development stage of HIPA-NET, and therefore CCITT X.25 protocol is adopted as the basic protocol of HIPA-NET.
In order to increase the usage rate of a leased circuit, the message storage facility in a packet switching node is used actively in this system. In addition, new service facilities that are performed by the use of the storage facility in the node are provided. A set of protocol is defined to carry out these new services. This protocol is placed on the upper protocol layer of X.25 level 3.
In this paper, system structure, protocol, confirmation of transmitted messages, etc. of this private packet network are described.
The following can be considered important motives for a firm wanting to have its own private communication network.
(1) When new communication services that are not provided by a public network are required.
(2) When a private network is expected to be more economical than a public network.
(3) When a high degree of message security is not insured by a public network.
Item (3) may be a special case. Items (1) and (2) are considered to be the main requirements. Recently, the need to construct private networks is being accelerated by the increase of new comnunication traffic such as facsimile communication, data communication, etc., and by the expansion of geographical activity area of a firm. A private communication network has to accommodate the above-mentioned requirements.
On the other hand, some new communication services are expected to be provided by a public network. And therefore, compatibility and interworking between private and public communication networks are not to be ignored.
From the point of view of message delay time, with regard to a terminal, communication facilities are classified as shown in Table 1.
Table 1 Classification of communications
|1||Real time communication||Telephone, Voice input. Voice response|
|2||Small delivery delay communication (less than 1 minute)||Conversational TSS, Conversational teletype. Urgent facsimile, Information search|
|3||Large delivery delay or on-demand output communication||Ordinary facsimile, mail box, telex|
In most communications, except telephone and on-line real time computer access, a transmission delay in a communication network from several minutes to several hours may be permitted. By multiplex use of a communication circuit by a number of communication facilities, the efficiency of the circuit usage is increased. And as a result, circuit cost per message is decreased. Such a scheme is possible when a relatively large transmission delay is allowed for each communication.
In order to increase the usage rate of a circuit, a message queue at the sending terminal may occur. Since the maximum queue size increases proportionally to the usage rate of the circuit, a relatively large buffer storage is required.
Sometimes, on-demand output of a received message at a receiving terminal is required instead of automatic output. Such requirement arises in the following cases.
(1) When a confidential message to a particular person is to be sent.
(2) When a message to be sent is input, regardless of status of a receiving terminal, for example, the receiving terminal is in power off state. Such a case is often encountered if a large time difference exists between a sending terminal and a receiving terminal.
In order to increase the usage rate of a circuit and to realize on-demand message output, it is essential to prepare a large message buffer storage at nodes in a communication network.
As mentioned in the previous sections, the basic requirements of a private communication network are as follows.
(1) Multiplexed use of the same communication circuits for various kinds of communication.
(2) Preparation of a large message buffer storage at the nodes in the network.
(3) On-demand output of a message at a receiving terminal.
(4) Ease of interworking with a public communication network.
Considering these requirements, the followings are adopted as the basis of HIPA-NET design.
(a) A packet switching network that accommodates CCITT X.25 protocol and other CCITT recommendations on circuit interfaces.
(b) A large message buffer storage at each packet switching node.
In addition, the Permanent Virtual Circuit (PVC) concept is adopted throughout the network so as to simplify the connection control. PVC is suitable for a private network since the total number of terminals is small compared with a public network, and destination terminals from any one terminal are limited.
Fig. 1 Model of Communication
A simplified system structure is illustrated in Fig. 1. PVC links are set between terminals (A) and node (1), node (1) and node (2), node (2) and terminal (B) respectively. A message is transmitted from terminal (A) to terminal (B) as follows.
(1) The message is transmitted from terminal (A) to the buffer storage in node (1). The buffer storage is regarded as a virtual destination terminal.
(2) The message is transmitted from node (l) to node (2).
(3) The message is transmitted from node (2) to terminal (B) automatically or on-demand from terminal (B).
Linkage between PVC (A, 1) and PVC (1, 2), PVC (1, 2) and PVC (2, B) is determined according to the content of a protocol header and PVC network table in each node. Other node(s) may be inserted between node (1) and node (2).
A concept of selective connection in a PVC network is shown in Fig. 2. A number of logical ports are provided to each terminal. By activating a particular logical port, a destination terminal may be identified selectively.
Fig. 2 Concept of PVC network
CCITT X.25 protocol is applied to HIPA-NET considering compatibility with public packet switching networks. In order to achieve mail type communication services. Transport Service Protocol (TSP) and Mail Service Protocol (MSP) are placed over level 3 of X.25. as illustrated in fig. 3.
(1) TSP protocol
In TSP, procedures of transfer control, data flow control and session control are defined. TSP is based on Hitachi's standard network architecture HNA, and applied to the communication between an HNA terminal and a cluster terminal. The cluster terminal is a virtual terminal implemented in a packet switching node.
(2) MSP protocol
MSP is a protocol for message transfer control and message delivery control that characterize the mail service facilities. The detail of MSP is described in the next section.
(3) Users' application protocol
Users' application protocol is defined for terminal-to-terminal communication. HIPA-NET is not involved in this protocol.
The following functions have to be provided in the network to carry out communication correctly and to administrate the network effectively.
(1) Indication of the destination in a message
(2) Indication of start and end of a message
(3) End-to-end confirmation of a message to be transmitted
(4) Detection of loss or stay of a message in the network
(5) Recognition of the status of a destination terminal
Table 2 MSP Commands and Responses
|Transmission control||Start of transmission||Acknowledgement|
|End of transmission||-|
|Message delimiter||Start of text||-|
|End of text||Acknowledgement/Send monitor information|
|Message management||Send monitor information||Reply to send monitor information|
|Receive monitor information||Acknowledgement|
|Initialization||Terminal set||Reply to terminal set|
|Network management||Congestion indication||-|
|Remote trouble status indication||-|
|Terminal status check||Reply to terminal status check|
In order to provide these functions, a set of protocol is defined. Commands and responses that are defined in MSP are summarized in Table 2. In addition to the protocol, the following facilities ,are provided in the network.
(1) Sequential numbering of send messages at each terminal
(2) Sequential numbering of receive messages at each terminal
(3) Drawing up and distribution of invoice lists among nodes
On-demand received message output is carried out in the following manner.
Multiple logical addresses are assigned to a terminal that is used for confidential message output. One logical address is assigned to ordinary automatic message output. Each logical address may be activated exclusively. Thus, logical addresses for confidential messages are usually inactive, and confidential messages are stored in the message buffer storage at the receiving node. A coded key is used to change the logical address of a terminal. By means of the coded key, a specific logical address is activated and messages destined for that address are output. Such a method may cause some problems on numbering plan if it is adapted to a public network. However, it is quite effective in a private network since restrictions on the numbering plan in a private network are less than in a public network. A priority indication bit is provided in the header to identify whether the message is an urgent one or an ordinary one. The urgent message is sent out from the buffer storage of each node in advance of ordinary messages.
In a communication network with data storage service or mail type service facilities, the data are held in network for a long period. Therefore, message numbering, confirmation of delivery and other operation management facilities have to be prepared in the network itself. The following facilities are provided in HIPA-NET.
(1) Management of message numbering
A send message number and a receive message number are given in order to each message at each sending node and receiving node respectively. These numbers are exchanged between the sending node and the receiving node to confirm right message delivery.
(2) Lists for management
Invoice lists, message number lists and not-delivered message lists are printed out at each station on operator's command input.
(3) Clock management
Real time clock of each node is adjusted from a network control center. In addition, start and end of summer time at any node and station is controlled automatically according to a program.
(4) Gathering of statistical data.
Traffic data, usage rate of each resource and other statistical data are gathered at each node and printed out periodically or by operator's command.
General idea of the message management is illustrated in Fig. 4. In this figure, communication between terminal A and terminal B is performed. A data terminal, named a management terminal, other than the communication terminal is prepared at each station for the sake of command input and list output. Message management procedure in Fig. 4 is as follows.
Fig. 4 Flow of Management Information
(1) Management procedure
(a) A send monitor message (S-MON) is printed out on the send terminal each time a message to be sent is received by a network node.
(b) A receive monitor message (R-MON) is printed out on the receive terminal each time a message is output to a destination terminal.
(c) A delivery confirmation message (DLV) is sent from the receiving node to the sending node and printed out by the management terminal in the message send station.
(2) Commands and lists for message management.
Lists in Table 3 are output on the management terminal in response to operator's command input.
Table 3 Commands and lists
|Invoice list (INVC)||A list of messages from a station to every other station. Sent to each station separately when an INVC command is input at the send station.|
|Last message list (LAST)||A list of total invoice and number of messages sent in a day. Printed out in response to LAST command input at the closing hour of a business day.|
|Not-delivered list (NDLV)||A list of message identification (destination address, send message number, etc.) staying in the network or lost in the network. Printed out in response to NDLV command input.|
|Check message (CHCK)||A message indicating whether a designated message is delivered or not. Printed out in response to CHCK command with the message identification.|
Contents of each message and list are summarized in Table 4.
Service facilities, system structure and protocol of an integrated service private packet network have been introduced. The private packet network named HIPA-NET has already been supplied to some firms, where facsimile, teletype and other messages are being transmitted around the world by means of international voice grade leased circuits.
Table 4 Contents of management messages and lists
|1||Send message number||o||o||o||o||o||o||o|
|2||Receive message number||x||o||x||x||x||x||x|
|3||Send station name||x||o||o||o||o||x||x|
|4||Receive station name||o||o||o||o||o||o||o|
|9||Command input time||x||x||x||o||o||o||o|
|10||Number of messages||x||x||x||o||o||o||o|
in the network
x Not included
 S. Yamaguchi et al, "Design of a private packet network", Proc. ICCC 78, pp 89 - 94 (1978)
Shoichiro Yanaguchi, engineer with Totsuka Works, Hitachi, Ltd., graduated from University of Nagoya in 1971. Since then, he has been working on development of data switching, especially packet switching.
Keisuke Tomaru graduated from Gumma University in 1955. He had been engaged in the research and development of electronic switching systems and computer communication systems in the Electrical Communication Laboratories of NTT. In 1978, he joined Hitachi, Ltd. At present, he is in charge of new communication systems development.
Takao Kato, Senior Engineer of Hitachi Ltd., graduated from University of Tokyo in 1965 and joined Hitachi, Ltd. Since then, he has been working on design and development of telephone switching and data switching.