Inside packet sampling techniques: exploring modularity to enhance network measurements

Carvalho

Bispo

et al. 2019

Int J Communication

Self Cite

Currently deployed in a wide variety of applicational scenarios, wireless sensor networks (WSNs) are typically a resource-constrained infrastructure.Consequently, characteristics such as WSN adaptability, low-overhead, and low-energy consumption are particularly relevant in dynamic and autonomous sensing environments where the measuring requirements change and human intervention is not viable. To tackle this issue, this article proposes e-LiteSense as an adaptive, energy-aware sensing solution for WSNs, capable of auto-regulate how data are sensed, adjusting it to each applicational scenario. The proposed adaptive scheme is able to maintain the sensing accuracy of the physical phenomena, while reducing the overall process overhead. In this way, the adaptive algorithm relies on low-complexity rules to establish the sensing frequency weighting the recent drifts of the physical parameter and the levels of remaining energy in the sensor. Using datasets from WSN operational scenarios, we prove e-LiteSense effectiveness in self-regulating data sensing accurately through a low-overhead process where the WSN energy levels are preserved. This constitutes a step-forward for implementing self-adaptive energy-aware data sensing in dynamic WSN environments. KEYWORDSadaptive sensing, data collection, energy-aware sensing, wireless sensor networks INTRODUCTIONAs a key enabler of the Internet of Things (IoT) and Smart Cities paradigms, wireless sensor networks (WSNs) are currently a technology capturing much attention, both from the academic community and industry. Sustained by the development of multifunctional low-cost wireless sensors, WSNs are applied in a large range of scenarios, many of then requiring operation without human intervention. 1,2 This versatility and difficult maintenance urge for mechanisms of self-management and power saving in order to allow a cost-effective adaptation of the sensing process to the application area, while optimising WSN lifetime. 3 Reducing energy consumption of sensor nodes for improving WSN lifetime has been a major topic of research in the last decade, being identified three major subsystems impacting on energy consumption-communication, sensing, and processing. The communication subsystem revealed to be the most demanding, even for devices able of energy harvesting. 4 This motivated solutions such as data gathering aggregation in order to reduce the number of transmission events.However, several works revealed that, depending on the operational scenario, acquiring and processing data may be more demanding than communicating. 5,6 This suggests that a versatile data gathering process should avoid acquiring redundant information, even when data aggregation strategies are in place.Int J Commun Syst. 2020;33:e4153.wileyonlinelibrary.com/journal/dac

Section: Related Workmentioning

confidence: 99%

e‐LiteSense: Self‐adaptive energy‐aware data sensing in WSN environments

Carvalho

Bispo

et al. 2019

Int J Communication

Self Cite

“…The concepts of sampling are represented in Sampling techniques have several characteristics that identify them and may be grouped by their granularity, selection scheme and selection trigger. The definition of each of these components is as follows [38]:…”

Section: Sampling In Monitoring Systemsmentioning

confidence: 99%

Deploying Time-based Sampling Techniques in Software-Defined Networking

Teixeira

2018 26th International Conference on Software, Telecommunications and Computer Networks (SoftCOM)

Lima

2018

Self Cite

“…Due to the constant increase in data volumes, it is also essential to enhance the monitoring process without compromising its efficiency. For this purpose, the use of traffic sampling techniques is mandatory to enable capturing the behaviour of services and networks resorting uniquely to a subset of traffic [1].…”

Section: Introductionmentioning

confidence: 99%

An Ontology-Based Recommendation System for Context-Aware Network Monitoring

Advances in Intelligent Systems and Computing

Carvalho

Lima

et al. 2019

Self Cite

Current network management systems urge for a context-aware perspective of the provided network services and the underlying infrastructure usage. This need results from the heterogeneity of services and technologies in place, and from the massive traffic volumes traversing today's networks. To reduce complexity and improve interoperability, monitoring systems need to be flexible, context-aware, and able to self-configure measurement points (MPs) according to network monitoring tasks requirements. In addition, the use of sampling techniques in MPs to reduce the amount of traffic collected, analysed and stored has become mandatory and, currently, distinct sampling schemes are available for use in operational environments. In this context, the main objective of this paper is the ontological definition of measurement requirements and components in sampling-based monitoring environments, with the aim of supporting an expert recommendation system able to understand context and identify the appropriate configuration rules to apply to a selection of MPs. In this way, the ontology, defining management needs, network measurement topology and sampling techniques, is described and explored considering several network management activities. A use case focusing on traffic accounting as monitoring task is also provided, demonstrating the expressiveness of the ontology and the role of the recommendation system in assisting context-aware network monitoring based on traffic sampling.