Engineering an extensible model for a public transport journey planning system

The research described in this thesis examined 1) the relevant issues of pre-trip public transport journey planning information; 2) the requirements for an extensible journey planning system; and 3) the necessary criteria from which to evaluate future, national, prototypes and models. It was found that, whilst comprehensive journey information is increasingly available via the Internet, it is not available in a coherent form and often requires the traveller to combine information from several sources in order to be able to decide on an integrated trip. The requirements necessary to allow a single entry point into a journey planning system were determined, a model developed, and a prototype of this model demonstrated using a protocol based on eXtended Markup Language (XML). In comparison to a warehouse model, a distributed data model and journey planning system design has the advantages of allowing expansion, both in geographical coverage and in flexibility of the protocol, in a relatively simple and incremental manner. Responses from individual components ranged from 0.05 seconds to 0.1 seconds for request for points, and 0.1 and 0.8 seconds for the next journey, given a database measuring about 10 megabytes. Requests for the whole day ranged from 0.9 seconds to just under three seconds. Network overheads constituted one-tenth of these response times. Application processes enabling extensibility were an order of magnitude smaller than these network overheads. Trial journey requests with multiple modes, exchange points, and search engines provided response times under 30 seconds in serial operation, and less than half that in parallel. Thus overall response time of these trial journey requests is expected to be most affected by the serial or parallel operation of requests, the speed of the particular search engines, and the quantity of exchange points. Further progress will be a function of hardware, software, and network limitations, availability of more granular (i.e. intermediate stop point) schedule data, and improved data linkages through geographic relationships.