Thierry Desprats

INCOME : multi-scale context management for the internet of things

Nowadays, context management solutions in ambient networks are well-known. However, with the IoT paradigm, ambient information is not anymore the only source of context. Context management solutions able to address multiple network scales ranging from ambient networks to the Internet of Things (IoT) are required. We present the INCOME project whose goal is to provide generic software and middleware components to ease the design and development of mass market context-aware applications built above the Internet of Things. By revisiting ambient intelligence (AmI) context management solutions for extending them to the IoT, INCOME allows to bridge the gap between these two very active research domains. In this landscape paper, we identify how INCOME plans to advance the state of the art and we briefly describe its scientific program which consists of three main tasks: (i) multi-scale context management, (ii) management of extrafunctional concerns (quality of context and privacy), and (iii) autonomous deployment of context management entities.

A generic multi-agent conceptual framework towards self-management

The high complexity of existing networks and services has led to the need to increase the autonomy of management solutions. This article presents a generic agent-based framework as a first step towards the conception of self-managed systems. This conceptual framework consists of groups of management agents coupled with managed resources and endowed with specific management skills. These groups of agents share a common representation of the managed environment which is specified with a dedicated model. From the analysis of this model, local and social agent behaviors are automatically deduced, thus participating in the overall autonomy of the management framework. An instance of this multi-agent framework is presented through an automatic intrusion response system

A survey on addressing privacy together with quality of context for context management in the Internet of Things

Making the Internet of Things (IoT) a reality will contribute to extend the context-aware ability of numerous sensitive applications. We can foresee that the context of users will include not only their own spatio-temporal conditions but also those of the things situated in their ambient environment and at the same time, thanks to the IoT, those that are located in other remote spaces. Consequently, next-generation context managers have to interact with the IoT underlying technologies and must, even more than before, address both privacy and quality of context (QoC) requirements. In this article, we show that the notions of privacy and QoC are intimately related and sometimes contradictory and survey the recent works addressing them. Current solutions usually consider only one notion, and very few of them started to bridge privacy and QoC. We identify some of the remaining challenges that next-generation context managers have to deal with to favour users’ acceptability by providing both the optimal QoC level and the appropriate privacy protection.

Enhancing context data distribution for the internet of things using qoc-awareness and attribute-based access control

The Internet of Things (IoT) enables producers of context data like sensors to interact with remote consumers of context data like smart pervasive applications in an entirely decoupled way. However, two important issues are faced by context data distribution, namely providing context information with a sufficient level of quality—i.e. quality of context (QoC)—while preserving the privacy of context owners. This article presents the solutions provided by the INCOME middleware framework for addressing these two potentially contradictory issues while hiding the complexity of context data distribution in heterogeneous and large-scale environments. Context producers and consumers not only express their needs in context contracts but also the guarantees they are ready to fulfil. These contracts are then translated into advertisement and subscription filters to determine how to distribute context data. Our experiments on a first open source prototype show that QoC-based filtering and privacy protection using attributed-based access control can be performed at a reasonable cost.

MuSCa: A multiscale characterization framework for complex distributed systems

Nowadays, complex systems are distributed over several levels of Information and Communications Technology (ICT) infrastructures. They may involve very small devices such as sensors and RFID, but also powerful systems such as Cloud computers and knowledge bases, as well as intermediate devices such as smartphones and personal computers. These systems are sometimes referred to as multiscale systems. The word “multiscale” may qualify various distributed systems according to different viewpoints such as their geographic dispersion, the networks they are deployed on, or their users organizations. For one entity of the multiscale system, communication technologies, non-functional properties (for persistence or security purpose) or architectures to be favored may vary from one scale to another. Moreover, ad hoc architecture of such complex systems are costly and non-sustainable. In this paper, we propose a scale-awareness framework, called MuSCa. This framework includes a characterization process based on the concepts of viewpoints, dimensions and scales. These concepts constitute the core of a dedicated metamodel. The proposed framework allows multiscale software designers to share a taxonomy for qualifying their own system. At system design time, the result of such a qualification is a model from which the framework produces scale-awareness artifacts. As an illustration of this model-driven approach, we show how multiscale probes are generated to provide multiscale components with an embedded scale-awareness ability.

From Ambient Sensing to IoT-based Context Computing: An Open Framework for End to End QoC Management

Quality of Context (QoC) awareness is recognized as a key point for the success of context-aware computing. At the time where the combination of the Internet of Things, Cloud Computing, and Ambient Intelligence paradigms offer together new opportunities for managing richer context data, the next generation of Distributed Context Managers (DCM) is facing new challenges concerning QoC management. This paper presents our model-driven QoCIM framework. QoCIM is the acronym for Quality of Context Information Model. We show how it can help application developers to manage the whole QoC life-cycle by providing genericity, openness and uniformity. Its usages are illustrated, both at design time and at runtime, in the case of an urban pollution context- and QoC-aware scenario.

Dealing with Stable Environmental Conditions in XACML Systems

XACML (extensible Access Control Markup Language) is an XML-based language for access control that has been standardized in OASLS. In this language, any entities involved in access control (i.e. users, resources, actions and environment) are specified by a set of attributes. This specification also includes the description of an architecture that explains how the policy decision point (PDP) retrieves the needed attributes values when it evaluates the policy to take its authorization decision. In this paper, we show this approach for getting the attributes values is a bottleneck to the performance of the authorization decision-making-process for attributes whose process for retrieving the value is long and the changing of its value doesnt impact the policy frequently. Thus, we propose an improvement of the XACML architecture in order to accelerate the decision-making-process when PDP has to treat such kind of attributes.

COBALT: an architecture for intelligent agent-based management

The ever increasing complexity of network and system management has lead to some changes in the organisation of management: we have moved to a more distributed approach. This approach uses two or three management levels, mainly management agents (SNMP agents) and a management station, but we forecast that this distribution trend will continue, towards a fully distributed solution. We call this solution intelligent agent-based management. In this context, cooperation between individual management agents is essential for them to solve complex management problems. We focus our contribution on the architecture which allows this cooperation. Agents must exchange tasks and information (like any management agent), but also functionality. Messages may be of the type query/response or knowledge exchange and may be sent from one agent to another, to several others or to all. We have examined protocols that could suit communication between our intelligent agents, but none entirely fulfils our requirements. KeML (Knowledge Query and Manipulation Language) was chosen as our communication language and CORBA as our transport language. This combination, named COBALT, constitutes the base of our architecture for intelligent agent-based management.

Composing data and control functions to ease virtual networks programmability

This paper presents a new domain specific language, called AirNet, to design and control virtual networks. The central feature of this language is to rely on network abstractions in order to spare operators the trouble of dealing with the complex and dynamic nature of the physical infrastructure. One novelty of this language is to integrate a network abstraction model that offers a clear separation between simple transport functions and advanced network services. These services are classified into three main categories: static control functions, dynamic control functions and data functions. In addition, we provide an easy and elegant way for programming these functions using the decorator design pattern. The AirNet language is supported by a runtime system handling, in particular, the virtual-to-physical mapping. Despite the fact that it is still in a prototype stage, this runtime has been successfully tested on several use cases.

Towards a virtualization-based control language for SDN platforms

Software defined networking (SDN) approaches rely on control languages to programmatically express the desired network behavior. Several SDN control languages use network virtualization to abstract the complex and dynamic nature of the physical infrastructure. However, almost all these languages use the same network abstraction model, which we believe is not the most appropriate one for expressing flexible and reusable network control policies. This paper presents work in progress towards a new high-level virtualization-based control language for SDN platforms. The main novelty of this language is to integrate a network abstraction model that explicitly identifies two kinds of virtual units: i) Fabrics to abstract packet transport functions and ii) Edges to abstract richer networking functions. We believe that this approach will allow network administrators to easily express modular and reusable network control policies independently of the underlying infrastructure.