Abstract
The convergence between mobile telecommunications and the Future Internet opened the way for the development of innovative pervasive computing services. The self-improving Personal Smart Spaces (PSSs) are coupling next generation mobile communications facilities with the features provided by the static smart spaces to support a more ubiquitous, mobile, context-aware, and personalised smart space. Addressing the advanced requirements of PSSs regarding the establishment of a robust distributed context management framework is a challenging task. Evaluating such a system is not a straightforward process, especially when it is also based on comparative assessments of its performance, as various existing systems demonstrate different unique characteristics making the quantitative comparisons quite complex and difficult to accomplish. This chapter elaborates on a context modelling and management approach that is suitable for addressing the PSS requirements and provides experimental evaluation evidence regarding its performance.
TopIntroduction
Pervasive computing is the third major era in computer science and aims to improve human experience and quality of life through ambient intelligence and smart spaces. In such environments users are served by interconnected smart devices without explicit awareness of the underlying communications and computing technologies. Towards this scope, there have been various research initiatives aiming the design and realisation of smart spaces (Lin &Jeng, 2013; Wang et al., 2013; Tusor & Várkonyi-Kóczy, 2011; Spadacini et al., 2014) in homes, offices, universities, schools, hospitals, hotels, museums, and other private or public places, where various automation facilities support the users. In these cases, research has focused on developing techniques to support building automation, as well as mechanisms to adapt the behaviour of electronic devices, desktops and peripherals, etc. However, these services are bounded in fixed spaces and the provided functionality is only available in geographically limited environments resembling to independent “islands” of pervasiveness in a sea of legacy service provisioning systems. On the other hand, services provided to mobile users are raising different and more challenging problems that need to be resolved. (Hansmann et al., 2003; Huang & Mangs, 2008)
The notion of self-improving Personal Smart Spaces (PSS), detailed in (Roussaki et al., 2012), couples the facilities offered by next generation mobile services and communications with the features provided by the static smart spaces. A PSS will provide to its owner a more ubiquitous and personalised smart space that is able to follow the user wherever he/she goes. Each PSS consists of multiple devices, both mobile and fixed, owned and administered by a single user. In addition, it facilitates interactions with other PSSs, it is self-improving and is capable of pro-active behaviour. Intelligent and dynamic behaviour is realised based on a combination of context awareness and learning techniques. PSSs models and monitors their owner’s behaviour and environment in order to be able to adapt according to each time situation. Hence, context awareness is a crucial feature for the PSS’s overall functionality. A plethora of heterogeneous context sources provide data which need to be collected, disseminated, processed and managed in an efficient way. Various information related with user and device location, user profiles, movement patterns, user activities, time, network performance, temperature, ambient noise level, personal preferences are exploited by services in order to automatically adapt their behaviour based on the user situation, requirements and intentions. In (Roussaki et al., 2012) the characteristics of PSS’s context management component were presented. The component has been designed and implemented based on a set of requirements such as real-time control and management of the context sources; distribution of context information over heterogeneous networks and devices; inference of future or currently unavailable high level context information from raw context data based on various learning techniques; modelling, management, storage and processing of history of context data; support for privacy-aware access control facilities and secure distributed context event management mechanisms; etc.
The presented context management system has been tested in real situation environments with successful results. However, evaluating a context management system and providing a comparative assessment of its outcomes is not a straightforward process. Various context management systems demonstrate different unique characteristics making difficult the quantitative comparisons. In this chapter the evaluation process in terms of context data management performance is presented. More specifically, this chapter is structured as follows. The next section presents the privacy-aware context model that has been implemented in order to represent the context information utilised in PSS environments and the respective quality of context metadata captured. The description of the high-level distributed context management architecture follows. Subsequently, a literature review for context management performance evaluation approaches is presented. The following section elaborates on the experimental evaluation of the discussed context management system. Finally, conclusions are drawn and future plans are detailed.
Key Terms in this Chapter
Context Refinement: Context refinement can be any method that accesses available context information of any kind and refines/enriches it, with the right quality and relevance. It is any process that creates new knowledge based on available context information ( Frank et al., 2009 ).
Personal Smart Space: A Smart Space is a collection of devices and software components that support context awareness features and proactively adapt to the user’s preferences and requirements. A Personal Smart Space is a Smart Space owned by a single user that follows its owner as he or she moves.
Context Management: Context management includes the distributed context maintenance, access, update and synchronisation, as well as, the provision of a transparent interface to context handling, support of ad-hoc context exchange, real-time and non-real-time context handling ( Henricksen & Indulska, 2006 ).
Context Awareness: Context-awareness refers to the idea that computers can both sense, and react based on their environment. Devices may have information about the circumstances under which they are able to operate and based on rules, or an intelligent stimulus, react accordingly. Context-aware devices may also try to make assumptions about the user's current situation. In addition, context-aware systems are concerned with the acquisition of context (e.g. using sensors to perceive a situation), the abstraction and understanding of context (e.g. matching a perceived sensory stimulus to a context), and application behaviour based on the recognised context (e.g. triggering actions based on context) ( Schmidt, 2002 ).
Platform: A platform is a set of software components or hardware subsystems, with related technologies that provides a set of precise functionality defined via interfaces and usage patterns. Components supported by that platform can interact and execute without being aware for the details of how the functionality provided by the platform is implemented.
Database Management System: A system or software designed to manage a database, and run operations on the data requested by numerous clients.
Context Database: A navigational database that extends over all Peers of the PSS.
Context Inference: The process of making context information explicitly available from other context sources (called lower-level context in this regard). It is part of Context Refinement (i.e. of all processes that create new knowledge based on available context information) and consists in drawing conclusions from existing information.
Node: A node is a physical device or virtual machine instance, which is capable of executing deployed services.
Context: Context is any information that can be used to characterise the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves ( Dey, 2001 ). Not all context information that follows that definition must necessarily be stored by a context management subsystem. While a context management subsystem provides convenient access to context information for third party services or platform services, such services may directly obtain context information from context sources (e.g. for efficiency) if these sources allow such access.
Sensor: A sensor is a device that measures a physical quantity and converts it into a signal which can be read by an observer or by an instrument ( Zhao & Guibas, 2004 ).
History of Context: The term history of context (HoC) denotes the recorded past values of context information regarding users, devices, services or other entities that may be part of or interact with the PSS or is somehow related to it.
Database: A collection of records, or pieces of knowledge. It is managed by a Database Management System.
Context Inference Rule: A personalised rule that indicates how to process contextual input in order to obtain other context information, that is not known before. A Context Inference Rule in general can take any form, e.g. logical, conditional, probabilistic or a NN.
Quality of Context: Quality of Context (QoC) denotes a set of parameters, used to describe the quality of certain Context Information elements. Examples of QoC parameters include: Accuracy, Probability of Correctness, Reliability, Resolution, Timeliness, Frequency, Price, etc. ( Zimmer, 2004 )