We will exploit the technology developed in QUANTICOL to consider scenarios which might change the current provision of transport service in cities, either on a shorter or longer term. Taking the city of Edinburgh as our example, the variability in resident population within the city provides significant challenges for transport providers. Edinburgh is the second most visited tourist destination in the UK, and during the Edinburgh International Arts Festival which runs from July to August each year the city’s population is estimated to double (to approach 1 million). This variability places great demands on a public transport system. Other cities where tourism plays an important economic role face similar challenges when they host major arts, sporting, or cultural events and would benefit from modelling tools which help them to plan their transport provision.
Quantitative system analysis supported by well-founded formal methods has the potential to help to make difficult strategic decisions in way that is robustly supported by carefully-considered analysis. A significant progress indicator for the project will be seen if the project researchers are able to use the mathematical tools and languages developed in the project to model complex scenarios in transportation in smart cities and deliver robustly-defended analysis which will be able to perform a key strategic role in shaping decisions in diverting resources in order to cope with increased demand for transport services.
The overall process algebra model will be analysed—perhaps statically therefore in a computationally efficient manner—to automatically extract subspaces within which to treat parameter estimation as a subproblem in isolation. Here, the movement events into the subspace are interpreted as source events, i.e. external arrivals of new agents into the subspace; movement events out of the subspace become sink events, i.e. events that destroy agents. This projection down into smaller subproblems also contributes to making the parameter estimation problem easier because it reduces a higher dimension problem to one of lower dimension.
In current approaches to parameter estimation in modelling, the general methodology is to carry out model fitting in a monolithic fashion, i.e. by estimating all the parameters of a model with a given set of observations. In the QUANTICOL project we aim to exploit the compositional nature of our CAS-SCEL modelling language to perform parameter fitting as the solution of a set of smaller subproblems.