Logical Neighborhoods: A Programming Abstraction for Wireless Sensor Networks

Abstract

Abstract. Wireless sensor networks (WSNs) typically exploit a single base station for collecting data and coordinating activities. However, decentralized architectures are rapidly emerging, as witnessed by wireless sensor and actuator networks (WSANs), and in general by solutions involving multiple data sinks, heterogeneous nodes, and in-network coordination. These settings demand new programming abstractions to tame complexity without sacrificing efficiency. In this work we introduce the notion of logical neighborhood, which replaces the physical neighborhood provided by wireless broadcast with a higher-level, applicationdefined notion of proximity. The span of a logical neighborhood is specified declaratively based on the characteristics of nodes, along with requirements about communication costs. This paper presents the SPIDEY programming language for defining logical neighborhoods, and a routing strategy that efficiently supports the communication enabled by its programming constructs.

A Formal Approach to Autonomic Systems Programming: The SCEL Language

“…This, and other similar forms of communication adopted in publishsubscribe architectures, can be easily rendered in SCEL by exploiting attributes for registering components' interests and by using predicates on those attributes for disseminating data to the registered components. [Mottola and Picco 2006;Mottola and Picco 2012] introduces the concept of logical neighbourhoods and the SPIDEY declarative language for defining them. Logical neighbourhood replaces the physical neighbourhood -i.e., the set of nodes in the communication range of a given device-provided by wireless broadcast with a higher-level notion of proximity determined by applicative information.…”

“…The communication mechanism enabled by the notion of logical neighbourhood is similar to the SCEL's one: predicates can indeed be thought of as a way of singling out the logical neighbours of a given node according to the features indicated by the attributes used in the predicates themselves. However, in [Mottola and Picco 2006;Mottola and Picco 2012] neighbourhood relations are statically defined through templates, while SCEL allows processes to form and use new predicates on-demand. Moreover, the SPIDEY language is specific for Wireless Sensor Networks (WSNs), while SCEL constructs are aimed at coordinating a larger class of systems/applications.…”

“…Indeed, we have learnt from software engineering the importance of component-based design and of separation of concerns [McKinley et al 2004], from multi-agent systems the relevance of knowledge handling and of spatial representation [Rao 1996;Bordini et al 2005;Winikoff 2005;Bellifemine et al 2007;Dastani 2008], from middleware and network architectures the importance of flexibility in communication [Mottola and Picco 2006;Costa et al 2009;Mamei and Zambonelli 2009;Nordström et al 2009;Mottola and Picco 2012], from distributed systems' security the role of policies [NIST 2009], from actors and process algebras the importance of minimality and formality [Agha 1990;Milner 1989]. What…”

A software architecture for adaptive modular sensing systems

“…An example of such a hierarchy is shown in Figure 1, in which a logical three-dimensional measurement sensor is implemented. Another existing architecture that provides similar benefits to that of a logical sensor hierarchy is that of logical neighbourhoods [4,5]. In this architecture, sensor and actuator nodes are abstracted into uniform virtual nodes that may in turn be further abstracted into a composite collection termed a logical neighbourhood.…”

“…However, the coordinate spaces of all physically connected TIMs must be defined relative to the TIM in the composite entity designated the pose base. Since the pose of the remote module would have already been transformed such that it is defined in terms of the coordinate space of the pose base, the local pose may also be brought into the coordinate space of the pose base by multiplying it by the remote pose, as per Equation 4. This operation is valid even if the pose base is only indirectly connected to both the local and remote TIMs, due to the accumulative effect of multiplying transformation matrices in which all previously applied transformations are carried over into successive transformations.…”