26-10-2016, 10:54 AM
1461236434-AnilResource.docx (Size: 4.39 MB / Downloads: 5)
Abstract
WSNs (Wireless Sensor Networks) are enablers that are essential of the concepts of Internet Things that help in envisioning the world of objects' interactions. The objects in WSN are nothing but small computers where sensors are equipped with control in capabilities of wireless communications. The specifications of WSN nodes are driven by constraints of energy since batteries used by them and operating may be required through long durations. Consequently, the hardware of these nodes utilize low-power electronics and their respective operation is defined by the capabilities of communications and limited processing.
In this Research paper, the case study of a solar SD (smart grid) is considered, where every solar panel has been equipped with a node of WSN which may help in generating a real-time stream to a sink. The surveillance of a Real-time video or monitoring give us the examples of such generating applications.
The real-time traffic generated by WSN nodes demands from the network a service characterized by parameters such as delay, packet loss, and throughput. In particular, we focus this work on guaranteeing a maximum End-to-End Delay (EED) at the application layer for packets transported by the WSN. A packet will be considered useful if delivered at the destination within the expected maximum EED, and useless otherwise. The transmission of useless packets consumes processing and communications resources, and contributes negatively to the congestion of the system.
The thesis aims to enhance the WSN support for real-time applications and efficiently use the WSN resources, by exploring the hypothesis that potential useless data packets should not be transmitted by the source node. Therefore, the thesis provides two major contributions: 1) a real-time mechanism to estimate packet EED based on IPv6 Routing Protocol for Low-Power and Loss Networks (RPL); 2) a cross-layer admission control mechanism that decides if a packet should progress towards its destination, based on the EED estimation available in each network node.
The proposed EED estimation mechanism was evaluated and the results obtained reveal that internal processing delays of the nodes are significant and they should be considered in order to accurately forecast the packet EED; RPL was also found to be usable as the instrument for
i
enabling the distributed estimation of EED. The packet admission control mechanism was also evaluated and the results obtained show that it actively contributes to decrease the number of the useless packets in transit in the WSN, consequently increasing the number of useful packets received at the destination, and improving the energy efficiency of each node, particularly under high network loads.
Introduction
Recent advances in the scaling of electronic circuits and in providing them with the ability to interact with the world around, enabled the appearance of Micro-Electro Mechanical Systems (MEMS). MEMS technology combines very small computers with sensor and control capabilities. MEMS mass commercialization and distribution made it very cost-effective and suitable for multiple uses. The deployment of communications capabilities in MEMS fostered the appearance of new network architectures and their interconnection to the existing global Internet Protocol (IP) network leaded to the birth of the Internet of Things (IoT) concept. The IoT envisions a world of interaction and coordination between objects, with or without human intervention, for the creation of smart environments.
Wireless Sensor Network (WSN) architectures extend the IoT concept by giving wireless communications capabilities to MEMS. A WSN is composed of a large number of sensor nodes, where each node can be characterized as a very small computer with a wireless interface. These nodes generate data from their sensors, such as temperature, humidity, moisture, and pressure, among others, and forward this data towards a gateway node. The gateway node, in turn, connects these networks to the Internet, as shown in Figure 1.1. WSN applications are multifold in areas such as smart metering, health care, environmental sensing, home automation, sports and wellness.
The hardware of the sensor nodes in a WSN is designed with processing and communications constraints since these nodes have limited energy resources. Even though these hardware limitations exist, more recently new and more complex applications and services (e.g. audio and video streaming) are pushed to be supported by the WSNs, in order to foster the concept of the IoT. These initiatives create new challenges in networking research areas such as routing, management, quality of service and energy efficiency.
1.1 Scope and Motivation
This thesis was carried out in the scope of the SELF-organizing power management for Photo-Voltaic Power plants (SELF-PVP) project [1] that aimed to increase the efficiency of a photo voltaic power plant with approximately 200.000 solar panels distributed in an area of 250 hectares. The solar panels include sensor nodes that communicate with each other using a grid topology WSN as shown in Fig. 1.2. In this scenario, we aim to deploy real-time applications, such as monitoring or video surveillance, in a set of sensor nodes.
Real-time applications typically generate traffic flows with Quality of Service (QoS) requirements that can be defined in terms of delay, jitter or packet loss. In case these applications require strict delay boundaries from source to destination, their packets must be delivered to the destination application within an End-to-End Delay (EED) limit in order for the information to be considered useful. The packets delivered outside the defined EED limit will be considered useless and discarded by these applications at the destination.
In order to enhance the operation of these applications, and since WSN nodes have relevant
processing and communications constraints, we explore the idea that the WSN should avoid processing and transporting useless packets and use its full potential to maximize the number of delivered useful packets. Therefore, our research is oriented towards the enhancement of the performance of a grid WSN considering the application’s viewpoint, while taking into consideration the efficient use of the available resources.
1.2 Problem Statement
A real-time application is to be deployed on a grid WSN where each node has limited resources in terms of processing, communications and energy. The real-time application generates delay sensitive flows with data that is assumed to be useful for the destination only if it is received within a strict delay boundary, and useless otherwise. In order to enhance the support for this application, the WSN performance can be oriented to maximize the number of delivered useful packets. At the same time, since WSN nodes have relevant processing, transmission and energy constraints, they should avoid to process and transport the useless packets.
The main problem to address is that the usefulness of a packet is determined at its destination, and processing, transmission and energy resources have already been expended
to transport the packet. Since the destination application may not consider all received packets as useful, if we are able to identify, as soon as possible, which packets will likely miss the application delay deadlines and avoid their transmission to the network, an increase in network performance and energy efficiency is expected to be achieved.
1.3 Objective
The main objective of this thesis is to enhance the support of real-time applications in a grid WSN topology, as shown in Fig. 1.3. In this topology, each source node generates a delay sensitive data flow directed towards a central destination node.
In each source node, the chosen strategy is to preview the EED of each packet and, as earlier as possible, avoid packet transmissions when these are expected to not comply with the limits given by the application. In order to pursue this strategy, the research efforts are divided in two particular objectives:
• Provide an EED estimation mechanism to be deployed in a WSN with minimal impact on network performance;
• Provide a WSN admission control mechanism based on the EED estimation and intended to enhance network performance and foster energy efficiency.
1.4 Contributions
This thesis provides two main original contributions:
• Novel mechanism to estimate EED based on RPL routing protocol
• In order to preview if a packet will be delivered within the EED limit defined by the application, a novel EED estimation mechanism is proposed. Other delay estimation mechanisms are proposed in literature but some of them do not provide a real-time and per-packet delay estimation, while others introduce additional traffic in the WSN to provide estimations. The proposed EED estimation mechanism provides a real-time and per packet EED estimation using IPv6 Routing Protocol for Low-Power and Lossy Networks (RPL). RPL packets are used to feedback the EED delay of the previously sent packets to the source nodes, thus avoiding extra traffic in the WSN. Also, to enhance EED estimation accuracy, this proposal accounts not only with transmission delays but also with the in-node processing delays which are relevant in the context of the limited processing resources of the nodes. This contribution has been published in [2]. Also, a set of RPL modifications to enhance the accuracy of EED estimation were proposed, and published in [3]. In the context of the EED estimation mechanism and in order to enhance EED estimation when using multiple network loads, a delay accounting optimization procedure was also proposed, and published in [4].
• Novel cross-layer admission control mechanism based on the EED estimation
• In order to decide if a packet should be transmitted accordingly to their usefulness to the destination application, a novel cross-layer packet admission control mech- anism is proposed. The proposed admission control mechanism is distributed by the WSN nodes and it is responsible for the decisions to transmit or drop a packet according to the requirements defined by the application. Other admission control mechanisms are proposed in the literature but the novelty of the proposed mecha- nism is that it runs in a cross-layer operation mode involving the application and network layers, while implementing interfaces with the EED estimation mechanism and RPL routing protocol. This contribution has been accepted for publishing in [5].