QOS management strategies for IOT applications

Abstract

This dissertation aims at managing and fulfilling two crucial Quality of Service (QoS) requirements of Internet of Things (IoT) applications, namely, latency and energy efficiency. Indeed, significant research efforts have been made toward achieving low latency and higher energy efficiency for latency sensitive and energy constrained IoT applications. However, the existing solutions possess various limitations. For instance, the literature on latency sensitive applications primarily emphasized on selecting low latency protocol to achieve minimal latency; certainly selecting low latency protocol is important, however merely selecting low latency protocol does not guarantee minimal latency. For example, in Bluetooth Low Energy (BLE) (a low latency protocol chosen for latency analysis and minimization in this dissertation) transmission, when the advertising interval is set too large, the BLE gateway receiving the packets must wait for advertising packets to arrive, which increases the overall latency. Conversely, with a small advertising interval, the advertising packets become overpopulated, resulting in extensive collisions. This, in turn, raises the overall latency. This shows the importance of selecting transmission parameters (advertising interval in BLE). However, in this context, none of the existing works have provided simplified expressions to choose optimal parameters for minimizing latency. Moreover, the current solutions often assume homogeneous network scenarios; however, the majority of IoT applications, including the smart manufacturing industry, have a variety of sensors having different data sizes, likely to form a heterogeneous network. Likewise, the solutions for energy constrained applications have the following shortcomings; Firstly, the sensors are powered directly from the power supply which makes the sensor to consumes few µA’s of current even in the sleep mode. Secondly, utilizes global positioning system (GPS) for real time location estimation even when higher accuracies are not desired or high errors in real time locations can be tolerated. Since it is a known fact that GPS requires a significant energy budget for its operation, hence affects overall battery life significantly. Thirdly, the existing solutions transmit raw data, usually generated frequently, consequently increasing total transmission and causing high energy consumption. Besides, it transmits a fixed number of copies to overcome data loss issues. However, transmitting a fixed number of copies may not completely overcome packet loss and may also lead to redundant transmissions. Hence, to address the aforementioned issues, this dissertation first proposes a performance model to analyze latency considering the heterogeneous scenario. Based on the performance model, derives simplified analytical expressions for optimal parameter selection that ensures minimal latency. Moreover, based on these analytical expressions, this dissertation proposes algorithms that autonomously optimize the network and ensure minimal latency. The work considers two different scenarios, namely, sensor nodes-gateway communication and sensor nodes-gateway-user communication. In addition to the above, this analysis provides valuable insights for service providers aiming to establish networks with such requirements. Thereafter, this dissertation proposes three different strategies for increasing energy efficiency and battery life enhancement, namely: (i) battery life enhancement through effective hardware design and efficient utilization, (ii) battery life enhancement through hardware miniaturization, and (iii) battery life enhancement through aggregated data transmission. Battery life enhancement through effective hardware design and efficient utilization involves: (a) utilizing general purpose input output pins (GPIOs) to power sensors and (b) optimizing controller clock configuration. Further, battery life enhancement through hardware miniaturization involves replacing expensive and energy intensive GPS with a energy efficient received signal strength indicator (RSSI) based real time localization algorithm which estimates real time locations without incorporating additional positioning hardware, thus saving significant energy. Furthermore, battery life enhancement strategy aggregated data transmission involves: (a) thresholding method that reduces the total number of transmissions and saves a significant amount of energy by only transmitting parametric data over raw data, which is usually sensed and transmitted very frequently, and (b) analytical expression for selecting number of copies required to overcome the packet loss and redundant transmissions, thus saving significant energy. In addition to the above, this analysis provides a foundation for an IoT engineer to achieve higher energy efficiency and longer battery life.