QoS Predictions WSN

BASIC INFORMATION

Short name: QoS Predictions WSN
Long name: Multi-Parametric QoS Predictions in Wireless Sensor Networks

Company: OMSATS University Islamabad
Country: Wah Campus

Call: F4Fp-05 (see call details)
Proposal number: F4Fp-05-M08

SUMMARY REMARKS & TESTBEDS

With the Internet of Things (IoT) proliferating and Wireless Sensor Network (WSN) deployment escalating and providing diversified applications, Quality of Service (QoS) concerns see challenging frontiers. Conventional algorithmic solutions find it difficult to optimize even a single metric because the associated optimization problem is often NP-Hard and lack adaptivity as the networks evolve. Joint optimization of multiple metrics is even harder with significant trade-offs. In addition, the benefit of any new protocol(s) cannot be realized until they are strenuously standardized and incorporated into the hardware, incurring significant cost. Data-driven techniques, benefitting from state-of-the-art Machine Learning (ML) algorithms have been a breakthrough in designing intelligent and adaptive systems. ML has been applied to various aspects of WSN. However, predicting QoS metrics (including Delay (D), Packet Delivery Ratio (PDR), Signal Strength (SS), Throughput (THP), and Energy Consumption (EC)) have been a new addition to this kind of research. It is a widely observed phenomenon that the QoS metrics are influenced by various configurable communication and networking parameters like Transmission Power (TP), Packet Size (PS), MAC Protocol (MACP), Routing Protocol (RP) and many others.

We intend to learn the relationship of such parameters with the QoS metrics to design a system that has a better and practical promise to meet the QoS criteria. Using techniques like ML can help achieve the adaptive design of WSN that will learn from its own experience and adapt to ever-changing and challenging circumstances posed by diverse deployment and application scenarios. The proposed framework can fine-tune the communication stack parameters to achieve a predictable performance for diverse and conflicting QoS goals. Furthermore, these fine-grained tunings of parameters will cast an impact on QoS in IoT.

MATERIALS