ERRORS-A POSITIVE APPROACH (POLYNOMIAL DATA COMPRESSION)

The object of this thesis is to show that a deliberate consideration of the function of errors within a system can produce novel and useful results.

In the polynomial data compression system, cubic splines are applied, with the aid of an error controlled algorithm, in such a way as to produce not just compression, but much better fidelity of reproduction. This results in the reproduced waveform being free of visual discontinuities, which is not the case for the conventional zero and first order data compression systems.

The control of error is basic to this algorithm's operation.

The measurement of error is then applied to amplifiers, the measured error being applied in such a way as to combine the advantages of both negative feedback and feedforward.

Of particular interest is the filtering employed to ensure a better stability margin.

In a redesign of a known amplifier, due to Blomley, the deliberate generation of gross error is used to overcome certain problems in the original design with the aid of subsequent measurement and error manipulation.

Some experimental results are presented.

Then follows a discussion of the current 'Block' Railway Signalling System in which the problems of the present system are discussed and a much improved system is proposed, in outline.

This treats the problem of Railway Signalling as involving very reliable communications, as well as detecting errors as they arise.

The prompt detection of errors is central to the proposed scheme.