Energy aware optimization for low power radio technologies
- Autores: Ioana Cristina Suciu
- Directores de la Tesis: Xavier Vilajosana Guillén (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2020
- Idioma: español
- Tribunal Calificador de la Tesis: Nouha Oualha (presid.), Carles Gómez Montenegro (secret.), José López Vicario (voc.)
- Programa de doctorado: Programa de Doctorado en Ingeniería Telemática por la Universidad Politécnica de Catalunya
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
The explosive growth of IoT is pushing the market towards cheap, very low power devices with a strong focus on miniaturization, for applications such as in-body sensors, personal health monitoring and microrobots. Proposing procedures for energy efficiency in IoT is a difficult task, as it is a rapidly growing market comprised of many and very diverse product categories using technologies that are not stable, evolving at a high pace. The research in this field proposes solutions that go from physical layer optimization up to the network layer, and the sensor network designer has to select the techniques that are best for its application specific architecture and radio technology used. This work is focused on exploring new techniques for enhancing the energy efficiency and user experience of IoT networks. We divide the proposed techniques in frame and chip level optimization techniques, respectively. While the frame level techniques are meant to improve the performance of existing radio technologies, the chip level techniques aim at replacing them with crystal-free architectures.
The identified frame level techniques are the use of preamble authentication and packet fragmentation, advisable for Low Power Wide Area Networks (LPWANs), a technology that offers the lowest energy consumption per provided service, but is vulnerable in front of energy exhaustion attacks and does not perform well in dense networks. The use of authenticated preambles between the sensors and gateways becomes a defence mechanism against the battery draining intended by attackers. We show experimentally that this approach is able to reduce with 91% the effect of an exhaustion attack, increasing the device's lifetime from less than 0.24 years to 2.6 years. The experiments were conducted using Loadsensing sensor nodes, commercially used for critical infrastructure control and monitoring. Even if exemplified on LoRaWAN, the use of preamble authentication is extensible to any wireless protocol. The use of packet fragmentation despite the packet fits the frame, is shown to reduce the probability of collisions while the number of users in the duty-cycle restricted network increases. Using custom-made Matlab simulations, important goodput improvement was obtained with fragmentation, with higher impact in slower and denser networks. Using NS3 simulations, we showed that combining packet fragmentation with group NACK can increase the network reliability, while reducing the energy consumed for retransmissions, at the cost of adding small headers to each fragment. It is a strategy that proves to be effective in dense duty-cycle restricted networks only, where the headers overhead is negligible compared to the network traffic.
As a chip level technique, we consider using radios for communication that do not use external frequency references such as crystal oscillators. This would enable having all sensor's elements on a single piece of silicon, rendering it even ten times more energy efficient due to the compactness of the chip. The immediate consequence is the loss of communication accuracy and ability to easily switch communication channels. In this sense, we propose a sequence of frequency synchronization algorithms and phases that have to be respected by a crystal-free device so that it can be able to join a network by finding the beacon channel, synthesize all communication channels and then maintain their accuracy against temperature change. The proposed algorithms need no additional network overhead, as they are using the existing network signaling. The evaluation is made in simulations and experimentally on a prototype implementation of an IEEE802.15.4 crystal-free radio. While in simulations we are able to change to another communication channel with very good frequency accuracy, the results obtained experimentally show an initial accuracy slightly above 40ppm, which will be later corrected by the chip to be below 40 ppm.
