Lecture 1: 10/1/2002 - Introduction to course.
Today I introduced the course content and administration details. I then
covered a motivation for the course by looking at some very general
communications topics to set the context and some focus for the
Lecture 2: 11/1/2002 - Human Interests.
Before looking at concepts, we review how computer communications affect
everyday life by looking at the systems humans use for communication. I
covered how communications have developed both for computers and for humans
using media very familiar to us. I introduced some terms to be used widely
in the course and gave some quantitative information on some communications
requirements. I covered up to slide 25 of the handout for this first block
of two lectures.
Lecture 3: 17/1/2002 - Fundamental Limitations.
I continued from last time looking at types of media most used in computing
and put some figures on typical transfer rates. In looking at how
computers communicate I introduced the three main strands of the course:
Information sharing, Time and Space as the first thrust; Agreement and
Implementation as the second and Human Influences as the third. To
introduce some ideas about fundamental limitations, I presented some
basic Information theory, all that you should need. However there is
more detail on the specialist handout.
Lecture 4: 18/1/2002 - Sharing Information.
The key concept here is that of sharing of information. We
established the kinds of things where agreement was essential, mostly
about datatypes to be handled. I introduced the notion of a
virtual machine, an abstraction to provide general interfaces to
the communications applications. I started on Compression and
covered Huffman coding, ie. up to slide 27. I handed
out the first part of coursework. The preliminary exercises relate to
the main practical to be handed out at a later date.
Lecture 5: 24/1/2002 - Sharing Information.
Today I picked up at Run Length Encoding, slide 28, and got through to the
start of DES encryption, slide 45. I'll have to finish this off tomorrow.
Note that the methods covered here for compression are much used today and
I looked at some common examples. Error correction is normally done by
retransmitting rather than by trying to correct on receipt.
Lecture 6: 25/1/2002 - Sharing Information over Time.
I briefly covered Encryption. Note that encryption is not a coding technique
as are compression and redundancy. I started on Time issues and covered up
to slide 17 which leaves a lot to do next Thursday. In today's lecture the
theme was on how bits are communicated and, as usual, we looked at Agreement
and Implementation, the recurring structure. Please note that there is
always recommended reading for each lecture block. These refer to the course
text but look for the same topic in other recommended texts as an alternative.
Lecture 7: 31/1/2002 - Sharing Information over Time - Time Packages.
Covered from slide 18 to slide 42 and will finish off this section tomorrow.
The main theme for this lecture was the implementation of time periods and the
introduction of ticket-style fow control, sliding windows and simple handshaking.
The lecture introduced several examples of time packages and handshakes.
Lecture 8: 1/2/2002 - SSCOP - Service Specific Connection Oriented Protocol.
I finished off the section on time: TCP, HDLC and LAPB. Since SSCOP fits in
very neatly with the section on time, it seemed right to give an overview at
this lecture. SSCOP is the protocol, some parts of which have to be
implemented for the main practical handed out today. We looked at
the format of messages and the protocol elements. There are six slides on
the web page with a very general description. I distributed an extra
handout with extracts from the SSCOP specification. This handout contains
all that is required for the main exercise and is an essential component
of the exercise. Covered slides 1-6 of the Lecture 8+9 note.
Lecture 9: 7/2/2002 - Introduction to Sharing Information over Space.
Communication space is the final piece of the model. As usual
there is agreement and implementation. Agreement is required for identifiers
for computers and channels forming the communication space. Implementation
covered: filtering; switching; splitting and multiplexing. Networks are
conveniently represented as graphs and an abundance of graph algorithms are
available for design and analysis. An important idea introduced is that of
the Multipeer channel which must be capable of allowing all N computers
connected to a channel to broadcast to the others at the same time.
Lecture 10: 8/2/2002 - Message Broadcasting Networks.
The structure for our look at networks is to look at the fundamentals of
information, time, space, channel implementation and multiplexing. The
name Message Broadcasting Network has been synonymous with Local Area
Networks but some techniques have now moved up to Metropolitan Area
Networks and Wide Area Networks just as we'll find that Message Switching
Network technologies are moving down from WANs. I covered types of
multipeer channel, use of Hubs and types of multiplexing. The first of
several examples we'll look at was Ethernet.
Lecture 11: 14/2/2002 - Message Broadcasting Networks.
I continued with further examples of MBNs: Token Ring, FDDI and FDDI-II,
Token Bus and DQDB. The evolution of LANs seems to have turned full circle.
First we had a centre master computer surrounded by communicating computers
and/or devices to form a Star arrangement. Now we have hubs giving us much
the same Star topology again but this time the centre machine is a switch.
Lecture 12: 15/2/2002 - Message Switching Networks.
Today we started on MSNs as these are the most common implementation now.
They are generally WANs but are now moving into MANs and LANs. As usual we
looked at information, time, space, channel, and multiplexing issues. We
concentrated on channel routing issues and routing issues to do with
multiplexing of which distance vector and link state were
considered to be important. I didn't have time to cover X.25.
Lecture 13: 21/2/2002 - Inter-networking.
I picked up with X.25 and stressed the use of ATM cell relay.
There followed several examples of public switching network services of
which the most important to recall is B-ISDN. I didn't cover these slides,
29-36. You should read them and the recommended reading. The next sectio
was on inter-networks and more important. The internet is a special case,
in fact THE case, but
here I wanted to cover some general properties of all inter-networks.
The main point is that different networks are linked and so the switches
have rather complex tasks to perform to translate information, time and
space differences. The terms router, gateway and
bridge were defined.
Lecture 14: 22/2/2002 - Case Study #1.
The first of two case studies looks at a protocol for a specific purpose
- factory automation. The requirements are for a bussed system and a real-
time service. The MAP project made two contributions to communications.
1. the MMS message protocol and the token bus LAN. In fact it made very
many more than this to communucations and other areas. The token service is
used to guarantee real time working. The network drives general-purpose
and special purpose computers, the latter driving manufacturing devices
such as robots, guided vehicles, etc. MAP closely follows the ISO set of
standards and stacks protocols.
Lecture 15: 28/2/2002 - Case study #2.
The second case study looked at a network that students will be familar
with - electronic mail. It is an example of a network built on top of
another network - the Internet. E-mail has its own servers and relays
and so really is a new network. I talked about SMTP, MIME and POP-3 and
looked at an example of ESMTP in some detail. Students should try using
the Unix `mail' command using verbose `-v' mode. In general E-mail got
the usual treatment, looking at information, time, space, channel and
Lecture 16: 1/3/2002 - The Real World.
The real world works to standards. We need them. We have to agree on
all sorts of processes and procedures to ensure successful communication.
There are a number of standards bodies and we looked at the major ones.
There are bodies representing different interest groups and bodies for
particular technologies trying to promote their standards.
First of all is the ISO reference model, a 7-layer model which is to be
used as an abstract model for classification and guidance and not for
implementation. The top three layers could usefully be considered one.
We looked at the layers in some detail.
Lecture 17: 7/3/2002 - The Real World.
We continued with the Internet standards, now the most important influence
on communications. The standards are much simpler than the ISO ones and
the Internet reference model (of sorts) is a three or four layer model.
It depends on whether one splits the core Network and Transport layer.
TCP/IP defines the Internet and UDP was added later than TCP. A short
summary of ITU's and IEEE's influence was also covered. Of proprietary
standards, we looked at the features of IBM's SNA and its major influence
on the ISO model and the Data Link Control services and protocols. A
summary of AppleTalk and Novell layering was presented.
There is a handout, a list of major non-proprietary standards available
through the course webpage via the Week-by-Week summaries link, and this
is a very useful summary of services and protocols and how they fit into
the layers of the ISO model.
Earlier this week Irwin hosted a tutorial on the second practical to ensure
that all understand exactly what is required.
Lecture 18: 8/3/2002 - George Ross on our networks.
The presentation covered many of the topics introduced in the course and
were tied together with a historical review and progress towards our
current network and that to be introduced over the next three years starting
this summer. George explained why and how we evolved with reference
to problems and their solutions. One point to note is that here at
Edinburgh, we have always been in the forefront of LAN development,