Wireless sensor networks (WSNs) have been used to observe and monitor many environments for specific purposes and in many ways over the past few years. A number of operational trade-offs are possible when planning a WSN, influencing coverage, bandwidth, redundancy, lifetime, expandability, and so on. However, for systems in potentially hazardous locations or those experiencing restricted access, system unreliability tends to be the greatest operational concern. In the process of creating reliable WSNs for hazardous locations, it is highly desirable to ensure both an accurate and a reliable system design prior to deployment, and with as little unnecessary trade-off as possible. Especially as sensing systems become larger and more complex, and potential failure modes increase, this becomes more difficult to achieve. In an attempt to answer the question of reliability assurance, the authors investigate WSNs in the context of accurate and fast modelling of such networks. A comprehensive comparison of three modelling tools (ns-2, OPNET, and OMNeT++) is explored in this chapter, concluding that OMNeT++ is worthy of study as an alternative to the other two more established tools. As an illustration of the use of OMNeT++, two modelling schemes are simulated and compared against the theory to determine both bit-level correctness, but also to demonstrate ease of modelling and analysis.
TopIntroduction
Mobile ad hoc networks (MANET) are self-configuring networks of mobile devices intended for particular purposes related to communication, and usually associated with wireless multi-hop communication (Figure 1). Two fundamental research challenges exist related to MANETs: the aspect of self-configuration – which is essential since each device in a MANET is free moving autonomously, and the limited reach of wireless communication links. These ideas are shared between MANETs and wireless sensor networks (WSNs) in general.
To date, WSNs have been well used in many different application areas, including scientific experimentation, engineering validation, environmental monitoring, health-related monitoring, industrial sensing & control, and also numerous commercial applications. MANETs are starting to be applied across the same range of similar areas, experiencing similar issues, compounded by their mobile (and ad hoc placement) nature. When such systems operate in a real environment, system unreliability may become the principal concern. For such systems, we need to answer the following questions:
- 1.
Can we simulate such system to some degree of accuracy that will ensure high reliability in the field (with some trade-offs accepted, or course)?
- 2.
As sensing systems become larger and more complex, the “try it and hope it works well” approach becomes less likely to succeed; we need an assurance that a system will work, before it is deployed. Can we ensure this through modelling?
We propose the use of fast but accurate scenario modelling to answer these questions - and will describe the practical use of OMNeT++ in aiming to resolve answers to both of the above questions.
To easily and conveniently model WSNs deployments and their failure scenarios, a reliable simulation tool is required. The aim is obvious; to model and predict the system-level behaviour of the real world environment as precisely as possible. There are a number of network simulators related to sensor networks (as can be seen later). It is important, in order to effectively develop any protocols with the help of simulation, to know the benefits and weaknesses of the various simulators. This will aid us in choosing which simulators should be used in terms of flexibility, reliability, and budget.
We explore the features of the three most commonly used network simulator tools, which are ns-2, OPNET, and OMNeT++. Then we explore various related issues (i.e. availability of protocol and topology models, debugging & GUI environment, simulation comparisons, and open source & cost issues), to conclude that OMNeT++ is worthy of study. We restrict the discussion to the smaller, less expensive research tools, rather than the more professional, capable, and higher cost tools such as Qualnet. The endpoint is ideally a low cost, preferably open source, capable and fast tool with ease of use in planning WSN deployments for hazardous locations.
Within this chapter, we also simulate the concepts of simple wireless transmission, and built this step-by-step into a full WSN simulation using OMNeT++ as an illustration of the process. First, we will model a simple concept of lost and received packets with the condition of variable Bit Error Rate (BER). We then match this to analytical equations for BER, Packet Error Rate (PER), goodput and efficiency, and simulate various conditions. We the progress to a more advanced and realistic simulation, introducing the concept of Segmentation and Reassembly (SAR), defining a window size as a ‘dimension’ of a packet that is sent over the air, to explore various system trade offs. Next, we add an error control mechanism (inspired by TCP), with retries and reACKs schemes similar to the radio link protocol (RLP). Lastly we will consider overall system efficiency issues. Simulation results are compared against theory, and various trade-off issues explored briefly.