The new generation of personal mobile devices has significantly changed the way humans interact with one another, using powerful, programmable devices equipped with a range of quite advanced sensing capabilities such as GPS, cameras and accelerometers. Advances in sensor networking and the availability of devices enabled with low-cost sensors have led to the integration of sensors to the monitoring and instrumentation of the physical world. We already see this in a variety of economically vital sectors such as agriculture, transportation, healthcare, critical infrastructures etc. However, such sensors are scattered, belong to a variety of entities and their data can be of interest to many applications and other business actors. Thus, the exponential growth in data capture and data sharing capabilities is creating a compelling need for the development of new mechanisms to support a community-sensing environment, where multiple users, business and community applications interact with each other and share data, using economics.

Current research initiatives have focused on solving the technical challenges that the physical environment brings to the scientific foundations of networking and information technology, such as enabling low-power, low-bandwidth wireless sensor network communication, energy are resource management, etc. These works have enabled a new era where people participate in sensing, instrumenting and analyzing certain aspects of their environment and activities, turning people into producers of "personal data", which is termed as "people-centric sensing" or "community sensing ".

However, the fundamental economic issue of why should users share or exchange such information, that may be costly to them but beneficial to others has not been addressed. To study this, one also needs to decide upon the necessary technology that can facilitate such exchanges within a sustainable system that can also encompass large growth in scale and user population.

In current approaches, incentives for generating information are very crude: users are basically limited to only two options in their interaction with a service provider, either to ?opt-in? and allow the given application they also use at the same time to have access to all the data they collect, or to ?opt-out? and accept generic, potentially low quality, levels of service, or no service at all. This ?tit-for-tat? (i.e. reciprocal) approach is very restrictive and not efficient. For example, users are motivated to provide data only for applications they use at that time, and there is no economic notion of exchange when user A has access to data that is relevant to user B and vice versa.

Bridging this gap motivates our proposed approach and research. We believe that a fundamental problem is to understand the socioeconomic implications of the "community sensing" environment when multiple users and communities interact, in order to provide support for its growth and make it sustainable. The goal is to leverage this social phenomenon and the parallel technological advances into a new paradigm on how people interact with each other in a virtual space using the principles and tools of economics. In our approach, the users owning sensing devices are treated as economic agents: they incur cost when they collect and transmit information, but they benefit by using social group applications that are based on the collected information provided by other members of the community too. There are important incentive issues regarding information provisioning since there is cost in collecting, transmitting and storing information, and users may be reluctant to provide information. Users and social groups would benefit if they were provided with the right "incentives" and could thus trade sensor information in a fair and efficient way, improving the performance of their applications.

To give a concrete example where the sharing of personal data can be very useful in a community sensing environment, real-time information on traffic encountered and the progress made by car drivers is very useful to other drivers that are planning their route: such information allows drivers to avoid currently congested streets and therefore speed up their journey time. At the next level, such sharing of information, provided it takes place at a large scale, can lead to the more efficient utilization of available routes at rush times, by optimizing jointly the routes of several individual drivers thus balancing traffic. Therefore, sharing such information has both clear costs (the cost of the transmission, but also the loss in privacy that the sharing of information entails) and benefits (better route planning) for the user. Since users like to maximize their value from using the system while minimizing their cost, they would like to free ride on the information provided by others. This would reduce drastically the amount of total available information and hence the quality and the value of the applications to the users. Without the incentive mechanisms we propose, a driver that is not at this particular instance interested in route planning will have her sensors off, depriving the driver community of potentially useful information. If the system could remember her contributions and reward her by allowing access to the route planning application at some later time, she may be incentivized to contribute more information even when she does not have an immediate benefit from this action. Similarly, information exchange between users may be the efficient solution: the given driver is interested about parking, but at the given time she can collect information only useful for determining route congestion; other drivers have access to information relevant for parking at the location our driver is interested and are ready to enter the freeway and hence care for congestion information.

The goal of this project is to develop a novel system that promotes incentive compatibility in sensor information exchange between users and applications, with regard to spatial, temporal and other related characteristics. The success of our endeavour is based on the following:

  1. Optimizing information management. The system needs to decide quickly, given the currently available data (provided by the users who have agreed to participate), how to satisfy users? requests. This is very crucial given the huge information space and the dynamicity of the applications consuming data. Following the example of the previous paragraph, some users may be interested in obtaining information about traffic patterns from various streets in the centre of a city. Another example would be requests for collecting samples for temperature or other quantities related to environmental measurements near a given area, e.g. someone interested in computing the average temperature and CO2 emissions of a region may want to collect samples with these values such that all samples are within a small distance from each other. Hence, such questions boil down to optimization problems under various sorts of constraints. To satisfy such requests, we need to develop algorithms that find appropriate matchings between the available data and the requests. In some cases, depending on the nature of the requests, these algorithmic questions turn out to be NP-hard and efficient heuristics will need to be designed.

  2. Game theoretic considerations. Any attempt for participatory sensing needs to take into account the fact that each user is a selfish entity, caring only for her own net benefit. We cannot expect that users in general will be willing to share their resources and information collected without receiving anything in return. Hence, to ensure that people are willing to contribute data, the right incentives should be given to them so as to provide the information that is most useful to the applications and the social groups that are active and avoid the generation of information that is costly and less useful. This is a major concern in many other contexts that involve interactions of economic entities. One potential solution is to come up with pricing rules for compensating the participants. Hence, in our context, we will need to combine the algorithmic solutions outlined in the previous paragraph with appropriate pricing schemes and mechanisms (such as auctions) that determine the refund that a contributing user receives as well as the price that a user that is asking for data should pay. Complications may arise here by the fact that a request may be satisfiable by combinations of data from different contributing users and it may not be a priori clear how to split the total payment to each contributor.

  3. Supporting infrastructure. To realise all the above in a real-life environment we need to create and develop the appropriate methodologies that will support the creation, development, and deployment in software of our economically-driven data collection and trading framework. This will be the focus of a dedicated area in the project. We believe that integrating the operation of a participatory sensing software platform with the economic framework will allow for the development of powerful data sharing applications, because this approach will provide a powerful, well-understood and distributed mechanism to optimize the information sharing cost and benefit trade-offs arising in all social environments in a way accepted by the self-interest entities participating.