دانلود رایگان مقاله چارچوب تحویل داده برای شبکه حسگر اطلاعات در نظارت هوشمند

Abstract

Cognitive information-centric sensor networks represent a paradigm of wireless sensor networks in which sensory information is identified from the network using named-data, and elements of cognition are used to deliver information to the sink with quality that satisfies the end-user requirements. Specialized nodes called Local Cognitive Nodes (LCNs) implement knowledge representation, reasoning and learning as elements of cognition in the network. These LCNs identify user-requested sensory information, and establish data delivery paths to the sink by prioritizing Quality of Information (QoI) attributes (e.g., latency, reliability, and throughput) at each hop based on the network traffic type. Analytic Hierarchy Processing (AHP) is the reasoning tool used to identify these paths based on QoI-attribute priorities set by the user. From extensive simulations, parameters that can be controlled to improve the values of QoI attributes along each hop were identified, and performance of the AHP-based data-delivery technique was compared with two traditional data-centric techniques in terms of lifetime and QoI attribute performance. It was found that the use of cognition improves the number of successful transmissions to the sink by almost 30%, while closely adapting the data delivery paths to the QoI requirements of the user.

6. Conclusions

In this paper, we proposed a framework for cognitive informa- tion centric sensor networks that can be used to implement infor- mation-centric data delivery using elements of cognition, i.e. knowledge representation, and inference to advance data-centric 990 sensor networks to cognitive information-centric sensor networks. 991 These CICSNs are able to handle heterogeneous traffic flows in the network generated as a result of requests coming from multi- ple clients in SOM applications, while considering the QoI attribute 994 priorities for each traffic flow. From the simulations we were able to identify the number of sensor nodes that should be simulta- neously scheduled while gathering data, to ensure good quality data from the sensor nodes. Optimally choosing the number of simultaneously transmitting sensor nodes improves the average 999 throughput by about 85%, reliability by about 90% and reduces the latency by about 18% for a given value of offered load (1000 bits). The simulation-generated values were used in the next set of simulations that implemented AHP analysis to decide the best next-hop node that should be used for data delivery to the GCN. It was found that the network lasted for significantly more number of transmission rounds, and performed well in responding to varying traffic types and changing network topology, when it implemented cognitive routing decisions, when compared with traditional decision techniques. In our future work, we will enhancing the learning strategy, and implement cache replace- ment at LCNs to further exploit the cognitive node’s capabilities to improve network performance and prolong the network life- time, while meeting the end-user’s requirements.