Reliability enhancement of packet delivery in multi-hop wireless sensor network

Abstract

Wireless Sensor Networks are used to collect a large amount of data in the field and forward it to data collectors, which are also responsible to analyze such data for a particular purpose. However, the wireless communication is susceptible to noise and interference, which results in loss of information in the routes from the sensors to the data aggregators. In addition, the devices used in these networks have restrictions on their processing, storage, and communication capacity, which diminishes, even more, the reliability of the data delivery. Therefore, this work describes a new communication method named μNet, which aims to increase the delivery rate of packets in multi-hop WSN. Such approach is based on peer-to-peer control messages to ensure that the packet has been properly forwarded. Besides that, the μNet is designed to be compatible with constrained devices, reducing the size of headers of the protocols and the size of the network buffer. By comparing μNet with widely used WSN protocols, it is demonstrated that this approach results in better end-to-end reliability in networks susceptible to interference.

Introduction

Sensing systems with wireless communication are used for many purposes, as fire detection in forests, soil analysis, weather sensing for forecasting, and many others.Alongside, with the advancement of the Wireless Sensor Network (WNS) technology, applications focused on Smart Cities and Internet of Things have become promising, providing ways to improve and facilitate citizens’ life [1] [2]. For example, a WSN can be used to reduce the energy waste in streetlight systems with a sensor attached to the light poles, making possible to detect the motion in the area and selecting the best illuminating behavior to the cities [3], as described in the Fig. 1.

The WSN is comprised of a large number of devices, usually in hundreds or thousands of sensor, in order to acquire data from the environment. Therefore, such devices must be of low cost to fulfill the requirements of these applications, which implies in resources constraints, such as: the processing unit, available memory for storage and processing, energy for operation and communication bandwidth [1]. In addition, these networks are usually located in environments that have sources of eletromagnetic interference, which reduces the system reliability and functionality [2]. Due to the unique characteristics of WSN, several methods were proposed to improve the WSN reliability, such as: enhance the network protocol [4], [5], [6], [7], [8]; adjust the network buffer [9], [10]; alternate the radio transmission channels [11]; adding system redundancy [12]; and others.

Despite the effectiveness of these protocols to enhance the network throughput and reliability, most of the proposed approachs do not take into account the high interference probability in the communication medium, which is a common and consistent problem in Smart Cities applications. The WSN usually uses Industrial, Scientific and Medical (ISM) spectrum to perform its wireless communications, that allows its usability without a license in most of the countries. Despite being common to develop wireless projects in this band, it also implies that it will suffer from the overload of this spectrum by all open systems that use it, known as ISM interference. For instance, the IEEE 802.11 (Wi-Fi) and the IEEE 802.15.1 (Bluetooth) are wireless communication standards that are generally deployed in cities, which directly impacts in WSN communications as demonstrated in [13].

Therefore, in order to provide a reliable WSN communication method for urban areas, Smart Cities and Internet of Things applications, this work describes a new communication protocol for WSN named μNet. By using a peer-to-peer acknowledgment system, the μNet protocol can deliver its packets with a smaller network buffer, since it is possible to control when the emitter will drop its packets by the acknowledge message. Comparing the μNet with the standardized WSN protocols with simulations that vary the interference probability, it is shown that the μNet approach is suitable for WSN with resources constraints located in environments susceptible to interferences in the communication medium.