Abstract: Traditionally routing in computer networks has focused on finding paths along which data packets could be delivered to pre-identified destination nodes. Most existing routing protocols rely on the use of network addresses as unique node or group identifiers that are usually numeric and independent of any application semantics. The semantically-oblivious identification has forced network designers to incorporate resource/service discovery techniques at higher layers of the network stack, resulting in unnecessary overhead. While such overhead can be tolerated in high-speed wired networks, it significantly limits performance and network lifetime in wireless infrastructure-less networks with battery-powered resource-constrained devices like sensor networks. Moreover, sensor nodes are more naturally anonymous and therefore assigning unique identifiers to individual node limits network scalability and imposes significant overhead on resource management. In this paper, we propose associative routing as a class of routing protocols that enables dynamic semantically-rich descriptive identification of network resources and services. As such, associative routing presents a clear departure from most current network addressing schemes, eliminating the need for a separate phase of resource/service discovery. We hypothesize that since, in essence, resource discovery operates similarly to path discovery then both can be performed in a single phase, leading to significant reduction in traffic load and communication latency without any loss of generality. We also propose a framework for associative routing and present adaptive multi-criteria routing (AMCR) protocol as a realization of associative routing for sensor networks. AMCR exploits application-specific message semantics, represented as generic criteria, and adapts its operation according to observed traffic patterns. Analytical results demonstrate the effectiveness, efficiency, and scalability of AMCR.
Abstract: In this paper, we present Adaptive Multi-Criteria Routing (AMCR), a novel routing approach for large-scale shared Sensor-Actuator NETworks (SANETs). Although many routing protocols have been proposed for wireless sensor networks, most of these protocols are designed with a specific application class or load pattern in mind. Furthermore, disparate concerns are usually woven within these protocols for optimization purposes. This limits the ability of these protocols to efficiently handle all the routing needs of shared SANET environments that may have different applications with different communication requirements. A novelty of the AMCR design is that it provides a truly generic routing protocol capable of exploiting the message semantics and adapting itself to the observed application characteristics in order to support the efficient operation of shared SANET platforms. AMCR provides descriptive criteria-based addressing of resources, and allows applications to publish their own resource criteria. It also provides seamless support for unicast/multicast/anycast/broadcast communication patterns. AMCR supports multi-criteria indexes to efficiently route complex queries. AMCR exploits the application-specific message semantics in routing, by allowing applications to create routing indexes for the most frequently targeted criteria. We also propose a mechanism for automatic adaptation of AMCR criteria indexes according to observed traffic patterns. The scalability and efficiency of AMCR is shown using analytical evaluation.
Abstract: We are witnessing a rapid expansion in the adoption of networked sensor-actuator systems (NSAS) deployed in support of applications such as smart homes, health management, public safety, and emergency management. Many of these emerging applications require large-scale deployment of NSAS and often have dynamic application-specific mission and evolving quality-of-service (QoS) requirements that include timeliness, reliability, security and availability. The shared and federated use of NSAS resources, to achieve multi-application goals, is a key to cost effective NSAS industry. This necessitates the decoupling of the NSAS physical infrastructure from application provisioning, and protecting applications and infrastructure resources from threats. The failure of NSAS nodes, due to malicious or non-malicious conditions, represents a major threat to the turstworthiness of NSAS platforms. Applications should be able to survive individual failures of resource nodes and change their runtime structure while preserving its operational integrity. Furthermore, for sustainable operation, QoS provisioning must be interwoven with energy conservation as a core priority in NSAS platform design. The large-scale of such networks, their heterogeneous node capabilities, their highly dynamic topology, their resource challenged nodes with the subsequent need for node cooperation, and their likelihood of being deployed in inhospitable environments, pose formidable challenges for the construction of trustworthy, shared NSAS platforms. To support the transition of NSAS from a research-only topic to a cost-efficient commercial industry that brings NSAS products and technologies to market, there is a need for a system-aided engineering methodologies and processes that addresses the industrial activities required for the full life-cycle of NSAS applications starting from the initial design to the evolution as requirements or mission change.
Abstract: Sensor Actor NETworks (SANET) represent for component of ubiquitous service environments promising interesting solutions to a wide range of problems. Despite the obvious increase in the research activities proposing architectures and protocols for SANETs, we are still no where near the production of industrial-grade SANET software that can be relied upon for mission critical applications. The cost of programming,prohibitive due to the lack of industrial tools capable of realizing adaptive SANET software in a cost effective way. We envision next generation SANET environments as large-scale autonomous systems 1) deployed by multiple infrastructure providers, 2)running multiple applications and 3) providing ubiquitous services collaboratively to both stationary and mobile users. Software adaptability is essential due to both the fragile topology and the change of the SANET goal or mission in response to diverse events.The system should be able to change its structure and behavior at run-time while maintaining its integrity. We introduce Elastic SANET (ESANET) as an attempt to realize a scalable, flexible,cost-effective and dynamic computing infrastructure capable efficiently running multiple applications on top of SANET hard-ware resources. ESANE1 software systems are likely to evolve over time. Deployment of software in ESANET is a progressive first class runtime operation. We propose a role-oriented software architecture that abstracts ESANET environments as fields of collaboration between specialized nodes, clusters, and overlays. We expect such architecture to achieve massive scalability and resilience to topology and context changes. Our proposed architecture could increase the network lifetime, not only by promoting efficient operation, but also by defining a mechanism to allow the network to recover from the death of individual nodes by preserving stocks of unspecialized standby stem nodes in a minimal power mode. Finally, we present a nano-middleware architectu- re and show how an evolution capable ESANET can be bootstrapped.
Abstract: The layered Internet architecture that had long guided network design and protocol engineering was an “interconnection architecture” defining a framework for interconnecting networks rather than a model for generic network structuring and engineering. We claim that the approach of abstracting the network in terms of an internetwork hinders the thorough understanding of the network salient characteristics and emergent behavior resulting in impeding design evolution required to address extreme scale, heterogeneity, and complexity. This paper reports on our work in progress that aims to: 1) Investigate the problem space in terms of the factors and decisions that influenced the design and development of computer networks 2) Sketch the core principles for designing complex computer networks and 3) Propose a model and related framework for building evolvable, adaptable and self organizing networks We will adopt a bottom up strategy primarily focusing on the building unit of the network model, which we call the “network cell”. The model is inspired by natural complex systems. A network cell is intrinsically capable of specialization, adaptation and evolution. Subsequently, we propose CellNet a framework for evolvable network design. We outline scenarios for using the CellNet framework to enhance legacy Internet protocol stack.
