On Distributed Communications Series

VIII. The Multiplexing Station

I. Introduction

A method of interconnecting high-speed all-digital Switching Nodes to form a highly survivable network structure able to transmit large quantities of data has been described in other volumes in this series. But, up to this point, only a portion of the overall communication network--the transmission subsystem for relaying Message Blocks from Switching Nodes to other Switching Nodes--has been detailed.

In the present Memorandum we expand the system description and describe the means by which network users, or "subscribers," enter the network to reach other subscribers.

Definition and Purpose of the Multiplexing Station

The term "Multiplexing Station" is used to describe that equipment which concentrates traffic from a number of simultaneous users into a form suitable for transmission by the Switching Nodes. Multiplexing Stations can be built in a wide range of sizes and for a wide variety of input devices. However, the description here is limited to one having the following specific characteristics:

  1. The Multiplexing Station shall process information from as many as 1024 separate subscribers.
  2. The Multiplexing Station shall perform all housekeeping operations necessary to convert low-data-rate input binary streams from any subscriber into Message Blocks for transmission into the network.
  3. The Multiplexing Station shall provide the data storage buffering means to unpack Message Blocks into a synchronous binary stream necessary for use by the end-addressees.
  4. The Multiplexing Station shall contain the means to interpret, modify, and transmit signaling information to and from its connected subscribers.
  5. The Multiplexing Station will probably, but not necessarily, be a manned station and shall contain a means for receiving trouble messages originating from nearby Switching Nodes, and for ascertaining system malfunctions.
  6. The Multiplexing Station shall contain complete on-line cryptographic apparatus to permit automatic on-line, end-to-end encryption of data from all locally connected subscribers for transmission to any other network user.
  7. The cryptographic apparatus used and contained within the Multiplexing Station shall be suitable for simultaneous handling of both classified and unclassified traffic. (This is required because many lines feeding the Multiplexing Station could originate from cryptographic ally non-secure areas.) The cryptographic equipment shall contain safeguards against cipher-breaking attempts based upon the insertion of known traffic into the network.
  8. The Multiplexing Station shall tie into two or three separate Switching Nodes by separate routes. These may be simple time-division routes multiplexed upon the same routes used between various Switching Nodes. (Alternate routes are necessary because of the limited reliability of a single route, but these alternate routes need not all be simultaneously available.)
  9. Key network subscribers shall be tied into two or more Multiplexing Stations. The Multiplexing Stations can be designed so that in the event of a major failure, the most important network subscribers are group-transferred to an adjacent Station for entry into the network--but with a loss of the end-to-end cryptographic and error protection.
  10. The Multiplexing Station shall be capable of simultaneous assignment of as many as eight separate office codes, for purposes of tactical mobility and reliability.
The large hypothetical Multiplexing Station described herein is designed primarily to handle traffic such as would exist from several hundred major military installations. We will first examine the case in which all the input devices (teletypewriters, telephones, etc.), together with the Multiplexing Station, are within a single, cryptographically secure area. (We will later discuss the case in which the input devices are connected by lines remote from the Multiplexing Station.)

For ease of description, the circuits between the subscriber and his Multiplexing Stations are called lines; lines are generally low-data-rate circuits. Circuits between Switching Nodes and, generally, those circuits operating at 1.5-megabit/sec rates, shall be called links.

Interconnection Between Multiplexing Stations and Switching Nodes

Figure 1 shows a possible interconnection configuration of Multiplexing Stations and Switching Nodes. One might imagine Fig. 1 as being representative of a portion of a large national network, comprising about 400 Switching Nodes and 200 Multiplexing Stations. The system itself is designed to handle any number of Switching Nodes together with 1024 Multiplexing Stations. Multiplying 1024 stations by 1024 subscribers connected to each, implies a tentative design capacity of about one million simultaneous network subscribers. But, further expansion does not appear difficult, if it should become necessary. It should be noted that in Fig. 1 there are only about half as many Multiplexing Stations as Switching Nodes. In this illustration, each Multiplexing Station is connected by a primary route to its nearest Switching Node and has two alternate routes leading to adjacent Multiplexing Stations via adjacent Switching Nodes. The routes used to connect the Multiplexing Stations to the alternate Switching Nodes in this example follow the same routes as those between the Switching Nodes. It is not difficult to time-division multiplex an additional channel over the same link as that used to transfer Message Blocks between Switching Nodes. Thus, at slight additional cost, alternate routes from each Multiplexing Station to its (adjacent) emergency alternate Station can be provided by using the same links that interconnect the Switching Nodes.[1]

Since Fig. 1 is drawn in the form of a highly regular grid pattern, Fig. 2 has been included to remind the reader that the highly regular canonical form of the distributed network will, in reality, be distorted to provide additional close-in access points to certain critical sites. The spacing requirement between the Switching Nodes is that they shall be so separated geographically as to minimize the probability of destroying more than one Switching Node with a single weapon. As unmanned Switching Nodes are physically small, it is economically feasible to physically "harden" them to further reduce the spacing dimension. Thus, Switching Nodes will be closely spaced in those critical areas where it is necessary to allow many communications feeder links to enter the network.

The precise number of Switching Nodes and multiplexing tie-points and their geometry is not examined in this series. However, we believe it is possible to provide the requisite communication means into the network at a few highly critical points, with a survivability level comparable to that of the remainder of the network.

A second form of topographic network distortion occurs when we consider the "real-world" location of the subscribers; this is illustrated in Figs. 3 and 4. Although it is convenient to refer to distributed networks with words that imply an overall uniformity of connectivity, this need not be a restricting condition. The best locations for major communications stations are not at uniformly spaced points, since best routes for communications links are rarely straight lines. Hence, some Switching Nodes may have many links to adjacent Switching Nodes, others few; Fig. 3 is a more realistic view of the topography of a real-world network. Figure 4 shows that the set of Nodes of Fig. 3 can be squeezed to fit a uniform matrix. It might be helpful to visualize the network as being drawn on a sheet of rubber, which is then distorted, for ease of visualization and analysis, to give the appearance of a neat rectangular configuration. Most survivability estimates for perfect-switching distributed networks have been based upon the assumption of a uniform 18x18 array of 324 stations.[2]

Thus, there are no inherent restrictions to the topography of the network implied in the following discussion, although we will describe a system of uniform connectivity.

[1] The mini-cost microwave links described in OD-VI operate at 4.5-million bits/sec. Each may be split into several lower-data-rate channels, such as three 1.5-million-bit/sec channels. (ODC is an abbreviation of the series title; the number following refers to the particular volume within the series.)

[2] Edge effects, caused by the outside edge of the network being less survivable than the interior of the network due to lesser connectivity, were ignored; it has been found that such effects insert only a slightly pessimistic bias in the survivability estimates for the network. This is briefly described in Reliable Digital Communications Systems Using Unreliable Network Repeater Nodes, by Paul Baran, The RAND Corporation, P-1995, May 27, 1960.

Next chapter