Pervasive computing envisions environments where computers are blended into everyday objects in order to provide added-value services to people. This vision is set to become reality and is actually raising huge expectations from consumers and service providers alike. As a result, many smart devices and open platforms are now proposed and installed in our homes and workplaces. Unfortunately, because of a lack of standards as well as a lack of cooperation between providers, the deployment of innovative services in our living places is today rather anarchic. This results in highly heterogeneous and sometimes incoherent installations, leading to major issues in terms of service efficiency, software maintenance, and even aesthetics. The purpose of this project is then to define and develop architectures, models and algorithms enabling interoperability in pervasive environments. In this approach, an application developed on a given platform could use services provided by another one. This goal is very challenging. Amongst other things, it requires platforms to build and maintain knowledge about themselves, about the environment, and about the capabilities of other platforms. It also demands abilities to conduct interactions with others in order to build shared contextual information and to find out if remote services can be actually used. Finally, it calls for a development framework including these facilities and facilitating programming activity. Finally, it is to be noted that interoperability features must be achieved in an autonomic way. A major constraint in pervasive environments is that there is usually no local administrator. Most management operations must be undergone by the system itself. The project will be based on two major open platforms, PCOM from Mannheim University and iPojo/iCasa from Grenoble University. These two successful platforms, heavily used and cited by academics and industrial, already present advanced features required to conduct this project. They are dynamic, autonomic and able to integrate heterogeneous devices and applications. More research effort is however necessary to reach interoperability as envisioned in this project. Besides a literature review on interaction in Pervasive Computing, PCOM and iPojo will be used as concrete platforms for our research. We will explore interaction protocols and context models that enable interoperability in Pervasive Computing and implement these in an architecture bridging PCOM and iPojo/iCasa. The relevance of the project outputs will be assessed on applications pertaining to the health domain. Providing services enhancing the potential of the elderly and patients to live and cope at home is today a major challenge, both in economical and societal terms. Healthcare costs are booming and the aging of the population only reinforces this trend. In this context, advances in interoperable pervasive computing are very encouraging and exciting.

Across the OECD, more than one in two adults and one in six children are overweight or obese, with obesity rates predicted to rise in the coming years. Obesity and poor diet contribute to a range of health problems, which increases the burden on publicly provided health services. There are substantial inequalities in the prevalence of diet related disease; for example, less educated women are two to three times more likely to be overweight than those with more education. This has led many governments to introduce fiscal and regulatory policies aimed at improving nutrition. Common policies include: taxes to raise the price of unhealthy food and drink; restricting what can be advertised and when; and limiting the availability of fast foods. There is a growing evidence base on the impacts of these forms of intervention, but it remains far from complete. In order to determine which policies are likely to be effective, we need to understand how people choose what to eat and drink, and how firms set prices, advertising and their product offering. A key component of people's purchasing behaviour is how what they buy today is affected by what they bought yesterday, or whether they saw an advert for a product last week. This dependence in decisions and experiences across time affects not only how individuals select what to buy, but also how firms set prices and choose how to advertise. Jointly, these market interactions will determine the effects of any government intervention on what's bought, and therefore what's eaten, and, ultimately, on health outcomes. Through this research we will develop methods to study these interactions, and we will apply the framework that we develop to two applications of topical policy interest. First, how does advertising affect people's food and drink purchasing choices, and how does this affect how firms choose to advertise. Modelling these decisions is challenging because of the complex and dynamic nature of advertising. If someone sees an advert for a product last week, or last month, this could still affect their decision to buy today. How it affects their decision is also complicated - does an advert for Coca Cola lead them to buy a Coke rather than Pepsi, or does it make them more likely to purchase any soft drink? Firms' decisions about how to advertise are both complex and interact with each other. Our objective is to harness data and computational developments to extend the literature to allow us to model the complexity of advertising competition in food and drink markets. Once we have estimated how people respond to advertising and how firms advertise, we can use the model to simulate counterfactual policies. For example, what would be the effect on prices, purchases and advertising if there was a change in tax policy, or advertising of certain products is restricted? These are questions that are currently of high interest to policymakers. Our second application is to study people's fast food purchases; consumption of food outside the home, and fast food in particular, is a substantial source of calories and has been linked to weight gain in children. Relatively little is known about people's patterns of fast food consumption, mainly due to a lack of data covering these choices. We aim to fill this gap by exploiting novel data on out-of-home purchases, which has the additional advantage that it is linked to the household's grocery purchases. This will allow us to study whether people with a strong preference for fast food, or who are particularly responsive to price or advertising, are also people from households that purchase many or nutritionally poor quality calories. This is crucial in determining the efficacy of targeting policy at fast food consumption and determining which type of policy to use.