Sistema de Telemetría basado en redes WSN (Wireless Sensor Network) para el Internet de las Cosas (IoT)

Abstract

The Internet of Things (IoT) involves a wide variety of heterogeneous technologies and resource-constrained devices that interact with each other. Due to such constraints, IoT devices usually require lightweight protocols that optimize the use of resources and energy consumption. Among the di erent commercial IoT devices, Bluetooth and BLE based beacons, which broadcast periodically certain data packets to notify their presence, have experienced a remarkable growth, specially due to their application in indoor positioning systems. As a solution to this problem, this thesis contributes with a general architecture of telemetry, di erent scenarios to access the cloud computing, guidelines for the selection of components in the di erent domains of an IoT-based architecture and examples of implementation of a WSN through a testbed based on the use of smart sensors with Bluetooth Low Energy. In the example shown, an IoT gateway is developed on an Android smartphone. This gateway integrates an MQTT client and has the necessary functionalities to pack the data from the sensors and send them to the MQTT broker residing in a cloud computing. There is also a friendly interface available for a local user to monitor and control the WSN created. Some experiments were carried out to demonstrate the functionality, speed and stability of the proposed gateway, to be used in this and other application domains. This thesis proposes a family of protocols named LP4S that provides fast responses and enables plug-and-play mechanisms that allow IoT telemetry systems to discover new nodes and to describe and auto-register the sensors and actuators connected to a beacon. Thus, three protocols are de ned depending on the beacon hardware characteristics: LP4S-6 (for resource-constraint beacons), LP4S-X (for more powerful beacons) and LP4S-J (for beacons able to run complex rmware). In order to demonstrate the capabilities of the designed protocols, the most restrictive (LP4S-6) is tested after implementing it for a telemetry application in a beacon based on Eddystone (Google's open beacon format). Thus, the beacon speci cation is extended in order to increase its ability to manage unlimited sensors in a telemetry system without interfering in its normal operation with Eddystone frames. The performed experiments show the feasibility of the proposed solution and its superiority, in terms of latency and energy consumption, with respect to approaches based on Generic Attribute Pro le (GATT) when multiple users connect to a mote or in scenarios where latency is not a restriction, but where low-energy consumption is essential. This thesis also presents a Virtual Transducer Electronic Data Sheet (VTEDS) based framework for the development of intelligent sensor nodes with plug-and-play capabilities in order to contribute to the evolution of the Internet of Things (IoT) towards the Web of Things (WoT). It makes use of new lightweight protocols that allow the sensors to self-describe, auto-calibrate and auto-register. Such protocols enable the development of novel IoT solutions while guaranteeing low latency, low power consumption and the required QoS. In order to evaluate the performance of the system, it was tested when using Bluetooth Low Energy (BLE) and Ethernet-based smart sensors in di erent scenarios. Speci cally, user experience was quanti ed empirically (i.e., how fast the system show collected data to a user). The obtained results show that the proposed VTED architecture is really fast, being some smart sensors (located in Europe) able to self-register and self-con gure in a remote cloud (in South America) in less than 3 s and to display data to remote users in less than 2 s. Finally, technological trends for the development of intelligent sensors in the eld of the current Internet of things are shown, based on communication technologies and protocols, hardware, multi-protocol and multi-network devices, operating systems, on-line development environments and programming languages of the rmware of said sensor nodes. All the scienti c contributions of this thesis are re ected in the design of a precision agriculture project, to control the irrigation of a banana plantation in an automated way. The aforementioned project is still in execution, although progress in its implementation is shown in the last chapter.