Políticas

In highly dynamic and mobility-enabled computing environments, users can connect to different points of attachment and wireless portable devices. They can also roam and expect to maintain connectivity and service sessions. For example, a user might be connected with his/her smart phone via GPRS (General Packet Radio Service) while outdoor, and wish to attach to the IEEE 802.11 enterprise wireless network once indoor. Disconnections of users/devices are rather common operating modes that can occur either, voluntarily, to reduce connection costs and to save battery power, or accidentally due to the abrupt loss of wireless connectivity, for example, because the user enters a no-coverage area. The result is that dependability issues and the automatic management of temporary disconnections should be treated as central and crucial functions of any support solution. In addition, mobile computing environments tend to be open and exhibit a high degree of heterogeneity of both access devices, for example, screen size/resolution, computing power, memory, oper- ating system, supported software, and networking technologies, such as, IEEE 802.11b/g, Bluetooth, GSM, GPRS, and UMTS (Universal Mobile Telecommunica- tions System).

As already stated, the main thesis of this chapter is that context-based approaches can offer suitable novel guidelines to support the above environments. Context is a complex concept that has several defi nitions [1,2]; here, it is considered to be any information that can characterize the state or the activity of an entity as well as the environment where that entity operates. More specifi cally, it can be the full set of metadata describing the current execution session, including the user’s preferences, characteristics and roles, as well as the service profi les and security policies. In fact, the high degree of variability and heterogeneity of the targeted deployment environ- ments makes service management a very complex task. Supporting and personaliz- ing services require novel methodologies and tools to specify which management actions should be taken based on contextual information [3,4]. For instance, consider the case of an enterprise user visiting a client company where he/she would be pro- vided with services, such as a printing facility or a virtual tour of the building. Mobile enterprise users should be enabled to dynamically search and retrieve location- dependent resources/services that could be of interest and supported them in carry- ing their tasks in the current context. Simultaneously, the scope of their visibility is to be restricted to avoid unintentional disclosure of sensitive business information. Finally, they should be able to act as service providers in addition to being service clients, by making resources and/or functions hosted on their devices available to other allowed users. Given the high dynamicity of these environments, proper access control solutions are required to permit the secure interaction of mobile users wish- ing to interact by reciprocally sharing and exchanging resources.

The increasing complexity of network management and application logic in con- text-aware systems raises the demand for adequate support solutions. In particular, several kinds of functionality for the support of wireless enterprise applications need to be provided at development and execution times, including: (i) the inter- action and/or integration of possibly heterogeneous systems (such as networks and

resource management systems); (ii) the abstraction of underlying facility details so to hide network, operating system, and programming language heterogeneity; and (iii) the implementation of these abstractions in transparent application program- ming interfaces.

A middleware-level approach starts to be widely recognized as a suitable solution to offer a set of fl exible and reusable support services that facilitate the design, deployment, and execution of a distributed application by implementing the above- mentioned features. In particular, middleware solutions for wireless enterprise net- works should be context-dependent and support at least two functions: (i) to securely discover only resources/services of interest for requesting users and (ii) to securely access resources/services, even hosted on portable devices, in open and dynamic wireless environments. Conventional middleware was designed for static contexts and tend to hide low-level network details. In contrast, because the context of enter- prise wireless networks is extremely dynamic, the corresponding middleware solu- tions should allow the adaptation of mobile computing applications to frequent changes in the execution environment and consequently frequent modifi cations in the applicable context [5].

Thus, the middleware should meet the following requirements:

1. It should collect and represent context information, such as user location, application execution conditions and device status, at a high level of abstrac- tion, and propagate this information up to the application level.

2. It should provide means to specify and enforce context-dependent adapta- tion strategies for running applications, transparently from the point of view of the application logic (thus requiring no modifi cations). For instance, a service providing a virtual tour of a building should adapt the data to be visualized to the characteristics of client device display, ranging from a laptop to a smart phone.

Context-aware behaviors should be expressed at a high level of abstraction by separating context-dependent decisions from the specifi c application logic and its implementation [1,6,7]. This separation of concerns is essential to reduce the com- plexity of service development and management in highly dynamic environments and to favor rapid prototyping, run-time confi guration, and maintenance of applica- tion components. This is pushing toward the proposal of novel middleware supports to assist service developers/managers/providers [4,6–8].

One primary mechanism to fulfi ll the above goals is through the exploitation of metadata. Metadata can describe both the structure/meaning of the resources com- posing a system and the management operations to perform when given conditions apply, expressed at a high level of abstraction [9]. Middleware solutions based on metadata for representing both context and adaptation of service behavior can pro- vide a high level of abstraction and a clear separation between service management and application logic.

Among the different possible types of metadata, profi les and policies are par- ticularly suited to build context-aware middleware solutions. Profi les represent charac teristics, capabilities, and requirements of system components, such as users,

devices, and services. Policies express the choices for ruling the system behavior, in terms of the actions subjects that can or must perform on resources [10]. Profi les and policies are maintained completely separated from the details of system implementa- tion and are expressed at a high level of abstraction, usually in terms of a declarative specifi cation language, thus ensuring the separation of concerns between context- aware application management and the implementation of the application logic.

The design-time and run-time effectiveness of metadata adoption, that is, respec- tively, the suitability to accelerate/simplify the development process and the light- weight effi cient exploitation of metadata to tailor service provisioning dynamically, depend on the characteristics of both the chosen specifi cation language and the middleware support infrastructure. Recent literature in the fi eld suggests that

semantic technologies represent a valid option for metadata specifi cation and man-

agement [3,4,11]. Semantic technologies consist in a set of languages and frame- works for expressive knowledge representation and reasoning. The powerful representation capabilities of semantic languages can model complex context infor- mation and dynamically extend defi ned context models with additional concepts and properties. As a key feature, semantic languages allow the formal specifi cation of context models whose underlying semantics is unambiguously defi ned, thus facilitating the dynamic exchange of context knowledge among interacting entities, even those that do not statically know each other, without loss of meaning. Let us suppose, for example, that a user is looking for a printing service within a client company wireless network. It might happen that there exists a printing service, whose name is unknown to the interested user. In this case, it is crucial for success- ful discovery that the user is enabled to express his/her desired service functionality based on the meaning of his/her query rather than on a particular keyword. Semantic technologies also allow automated reasoning to infer additional and/or more com- plex knowledge from available context data. The ability to reason over context knowledge can be successfully exploited to build middleware solutions capable of recognizing context and taking appropriate management decisions based on it. In addition, the adoption of semantic languages for metadata specifi cation simplifi es metadata reuse in open wide-scale deployment scenarios and facilitates the analysis of potential confl icts and inconsistencies.