Subjects -> Computer Networks -> Lectures -> Lecture #14

Lecture 14: Point-to-Point Data Links

Network Technology -- Point-to-Point Data Links

These are the most basic data communications links, eg:

Virtually all computers support such data links, using their so-called serial ports. Typically, these ports are able to operate at any standard bit rate between 300 bps and 230400 bps.

The most common hardware interface standard used in such data links is called RS232: in fact, the term RS232 port is widely used as a synonym for serial port.

A connection may be "direct" or may use a modem.

Modems

A modem allows a point-to-point data link to be carried over the analog telephone system (PSTN, also called POTS, or Plain Old Telephone System), thus:

The modem can:

Dial the telephone number of another modem under program control, or answer the phone.
Negotiate a connection at the highest common bit rate that each modem can support -- current maximum is 56kbps.
perform"on the fly" data compression, giving higher "apparent" data rate.
perform (under some conditions) error detection and correction -- able to recover from some line errors due to noise, etc.

RS232 Interface

This interface was originally based on a 25 pin connector^[1]. In the following, signal names refer to the computer (or DTE) end of the link: for example, pin 2 is "Receive data" on a modem -- that is, an RS232 cable is wired "straight-through". Major signals defined in RS232 are:

The RS232 interface was designed to connect a modem (DCE) to a computer (DTE). To make a direct (DTE-DTE) link, a null modem cable (or null modem adapter) must normally be used. This is sometimes called a crossover cable, since pin 2 at one end is wired to pin 3 at the other, etc.

More modern alternatives to the ubiquitous RS232 interface include USB and firewire, although most modern systems use a built-in modem.

^[1] 9 pin connectors are also used, but there is no standard pin assignment for this -- the PC/AT pinout is the most common.

Asynchronous vs Synchronous Framing

All data links have to provide byte-level synchronisation of the sender and receiver.

In an asynchronous system, such as is normally used with RS232-interface modems, each character (or byte) sent on a data link is prefixed with an extra start bit and has a stop bit appended, thus:

In synchronous systems, a whole block or frame of data is sent as a single unit, with a synchronising preamble of a few bytes. This has much lower overhead, and is by far the most common in "leased-line" data links, see later.

Integrated Services Digital Network (ISDN)

ISDN provides an all-digital replacement for the PSTN. ISDN is widely available in Australia from Telstra Corporation (although not from other telcos), and most Telstra customers now have the option of choosing it instead of PSTN.

The basic-rate ISDN service provides, conceptually:

An ISDN service is integrated in the sense that each B channel can be used to carry either a digitised voice (or video, FAX, etc) call, or a 64 kbps data call. The D channel is normally used for signalling (call setup, etc), but can be used as a 16 kbps data link under some circumstances. Although....

Note that in Australia, the ISDN interface is owned by the telecommunications provider (or Telco), eg Telstra. This is not the case in some other countries.

ISDN System Diagram

An ISDN service typically has the following physical components:

The S-Bus is a kind of dedicated LAN. It uses 4 (or 6) wires, and can extend between 100 and 600 metres from the NT1. A Terminal Adapter (TA) is needed for DTEs without ISDN interface.

Charging Policies

There is probably no more volatile marketplace at present than that for telecommunications services -- in many sectors (eg mobile communications), competition has resulted in large price reductions. This has, in general, not been observed in the "fixed service" market, where established telco monopolies such as Telstra in Australia are able to maintain comparatively high prices.

In general, PSTN and ISDN services are charged in a similar way:

Installation (or connection) cost, usually between $200 and $500 per service in Australia, depending on location, etc.
Annual (or monthly) "access" charge. Typically between $15 and $70 per month in Australia, depending on selected "plan".
Call costs, at widely varying rates depending on location, distance, "plan", etc.
In Australia, a strange disturbance in the political "force" means that local PSTN calls continue to be charged at a flat (per-call) rate, typically 25cents. This will change. Maybe... For ISDN, most calls are time charged, with differential rates for data and telephony calls -- the recent introduction of flat-rate voice calls on Telstra's new "Home Highway" ISDN service being an exception.

PSTN For Internet Access

Modems are nowadays most commonly used for dial-in Internet access, connecting to a dial-in router, a technology used by retail Internet Service Providers (see later) to deliver Internet access to isolated computer systems.

The usual configuration of a dial-in router is that hosts which connect via modem appear^[2] to be directly connected to the ISP's own LAN. Note also that the diagram is basically unchanged if an ISDN link is used: the modem at the customer premises is replaced with a TA, and the ISP has ISDN dial-in phone lines.

It's important to note in this diagram that the telecommunications service -- the phone line, sometimes called (in Australia) the "Basic Carriage Service" -- is provided by a "Telco": a telcommunications provider. This is conceptually separate from the service provided by the ISP: that of routing IP packets to and from the home user.

^[2] That is, the network (and subnet) parts of their IP address are the same as those of the computers on the ISP's internal LAN. The point-to-point link is "invisible" in terms of Internet routing.

The Point-to-Point Protocol (PPP)

In order to use a modem-based point-to-point link to carry IP packets, some kind of data link protocol is needed. Originally this was achieved using a very rudimentary "non-standard" protocol called SLIP. The current protocol for this purpose is PPP.

A PPP frame has the following structure:

The "FLAG" bytes delineate frame boundaries, and the data field contains an IP datagram.

When a PPP link is started, the protocol specifies a moderately complex set of "state transitions". Some of these states are:

Establishment: the PPP software exchanges Link Control Protocol messages, negotiating basic parameters for the link.
Authenticate: PPP includes facilities for verification of (eg) username/password pairs (PAP) to ensure that the link initiator is permitted to establish this data link.
Network: various essential network configuration variables are set during this phase, eg IP addresses and netmasks.

Other Internet Access Technologies

ISDN can potentially deliver 128kbps full-duplex (ie, symmetrical) ISDN service to the home user. For even higher speeds, typical recent technological developments include:

ADSL: Asymmetric Digital Subscriber Line provides a high-speed (in the low-Mbps range) data service over an existing phone line. The "Asymmetric" aspect is that the "downstream" data rate is usually much higher than the "upstream" rate -- an arangement which is suitable for most home users. It's also often possible to configure an "SDSL" link if preferred. This service is becoming increasingly available in Australia, and will probably dominate the market in future years.
Cable Modem: For areas which have an existing Hybrid Fibre-Coax (HFC) "Pay TV" infrastructure, the cable modem provides access to a shared high-speed medium (eg Optus@Home), in much the same (conceptual) way that traditional Ethernet works. Cable modems are limited by the fact that many areas do have a cable TV service.
Wireless Systems: A variety of systems have been proposed, none yet dominant. Watch this space...

Network Interconnects

We have already seen that the Internet consists of many networks/subnets) interconnected by routers.

Where a router is used to connect two geographically adjacent LANs (see last lecture), it is simply equipped with two appropriate LAN interfaces. Where the LANs to be connected are not physically near one another, it is normal for a point-to-point link to be used, thus:

See later for discussion of "leased line" links, commonly used in this context.

IP Addressing in Point-to-Point Links

Each of the LANs in the previous diagram is a distinct network (or subnet) for purposes of IP addressing and routing.

Not so obvious is the fact that the point-to-point link is also a network (or subnet), albeit one with only two hosts. For example, until recently the Bundoora and Bendigo campuses of La Trobe University were linked as follows:

Here we see that subnet 131.172.239.0 has been assigned to the point-to-point link between between the Bendigo and Bundoora campuses.

The tutorial for this lecture is Tutorial #14.
La Trobe Uni Logo

[Previous Lecture] [Lecture Index] [Next Lecture]