Section: Overall Objectives
Pervasive computing is based on the assumption of an integration of objects from our environment in the digital world. Such integration requires an understanding of the status of the real environment by computer systems to automatically be able to react to changes in that status. This is called context-awareness. Today, the necessary technology for the development of pervasive computing in the pioneering work of researcher Mark Weiser (https://en.wikipedia.org/wiki/Mark_Weiser ) are available: a ubiquitous Internet, short-range and low-energy communications, inexpensive sensors and actuators, and the expected democratization of the Internet of Things (IoT).
The objective of the TACOMA team is to build pervasive and context sensitive applications targeting Smart Spaces and considering all or some of the following five properties:
1. Data sovereignty and residency. The pervasive applications are backed by the capture of data in the environment. Most current solutions are sending the bulk of information collected on remote servers without the possibility for the user to control its spread and nature. Examples are smartwatches that transmit physiological user data to remote information systems to provide an overview of an individuals' physical activity. The collected data can also be used by insurance companies to customize their contracts according to the physical health of their customers. The final component of this scenario is nothing futuristic and illustrates the need to control the mass transfer of data generated by pervasive applications. Data sovereignty covers different aspects: strategic issues for the company that delivers the service using this data, strategic issues for companies wanting to use this service, legal constraints and the respect of user privacy.
The TACOMA research team considers four questions: What is the nature of the stored data? Where is it stored? Who holds it? Who can access it?
2. Acceptability. Perceptible pervasive applications rely on an instrumentation of the environment, particularly via sensors. This type of deployment is often hampered by two factors in the systems shown in the state of the art. First, a complex environmental instrumentation (e.g., an active ground to locate a user) results in unrealistic costs for pervasive systems. And the use of technologies considered too invasive by users (such as cameras or microphones) poses significant problems of acceptability.
The TACOMA research team considers that the instrumentation of the environment must be guided by a criteria of acceptability (operating somewhat intrusive sensors) and costs (using basic sensors that are very low cost, both economically and in regards to energy use).
3. Dynamism. The pervasive applications are deployed in environments subject to high variability. The resources accessible by the system depends on many factors: the mobility of the user, the presence of network connectivity, the surrounding objects, etc.
The TACOMA team apprehends this variability through two questions: How should a pervasive system react to the appearance/disappearance of resources (services/objects)? How can a pervasive system opportunistically exploit the available resources in the environment (for example, based on radio waves to power a sensor)?
4. Autonomy. The operation of connected objects is often based on access to a remote service through a centralized infrastructure or the cloud. We find this approach in the field of automation, for example, where smart thermostats capture data of the presence/absence or change in temperature. On this basis, a remote information system learns the habits of users and sends commands via the thermostat for heating the home. This strong dependence of a remote service poses problems in cost (the need to have access to infrastructure can undermine the economic model of a service), data sovereignty and acceptability .
The TACOMA team considers original and alternative approaches: autonomous systems that are integrated with the physical environment. Specifically, this autonomy is understood as a sort of strong independence of the application in relation to external infrastructure and remote information systems. This principle differs from more common approaches in the field of pervasive applications that use data to support processing only in a remote system.
5. Resilience. A pervasive application must deal with the volatility of the resources on which it is based. First, despite the ubiquity of the Internet, connectivity is extremely variable. For example, buildings have irregular radio coverage at the floor which could be problematic for the operation of a swarm of cleaning robots. Furthermore, objects may experience failures or breakdowns by giving false readings. It is necessary then to propose services that are equivalent or degraded to the user (more costly or with less functions, etc.).
The TACOMA research team believes in approaches that allow opportunistic resilience in using local resources. These approaches are based both on autonomous systems/structures (low infrastructure dependency) and the management of dynamism (elasticity of applications depending on available resources).