3122 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 54, NO. 6, DECEMBER 2007 Design and Implementation of an Educational Testbed for Experiencing With Industrial Communication Networks Lucia Lo Bello, Member, IEEE, Orazio Mirabella, Member, IEEE, and Antonino Raucea, Member, IEEE Abstract—This paper presents the design and implementation of an educational testbed developed for a course on industrial communication networks at the Engineering Faculty, University of Catania. The aim is to realize a platform capable of emulating various network configurations, thus enabling students to find out by themselves through practical experiments how different design choices, parameter settings, network configurations, and algorithms impact on the overall network performance. The test- bed comprises a number of basic components (hosts, routers, and access points) implemented on nodes equipped with operating systems and open source software, which together make up a modular system. Each router can be loaded with data flows and monitored over preestablished time windows so as to evaluate its performance in a wide range of operating conditions. The wireless part makes it possible to configure environments with different levels of noise and bandwidth utilization so as to emulate a broad spectrum of real operating environments. The testbed can be used via remote access through a web interface that not only allows the operating conditions to be configured but also permits real-time monitoring. Students can configure the testbed on the basis of the network they are studying and can measure its performance for evaluation purposes. Index Terms—Communication engineering education, com- puter networks, measurement, scheduling, wireless LAN. I. I NTRODUCTION O NE OF THE main difficulties encountered by students attending a course on industrial communication networks lies in the specific nature of these networks, which have to support various types of traffic flows with varying requirements in terms of bandwidth, delay, and time constraints [1], [2]. In addition, the interoperability of field bus systems with external systems and applications, which adds further complexity to the design of distributed process control systems, must be considered [3]. Another factor to be taken into account is the presence of both wired and wireless networks. As wireless networks feature quite different issues than wired ones [4]–[6] in order to understand how to properly design hybrid industrial communication networks, the adoption of new concepts and paradigms is needed. Manuscript received April 16, 2007; revised August 14, 2007. The authors are with the Department of Computer Engineering and Telecommunications, University of Catania, 95125 Catania, Italy (e-mail: llobello@diit.unict.it; omirabel@diit.unict.it; araucea@diit.unict.it). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIE.2007.907025 The traditional approach for students to learn and evaluate communication networks is based on the use of analytical or simulation-based tools. However, these tools only give an abstract view of the network behavior. As a result, it is very difficult for the students to assess the impact of the hardware or software on the overall performance taking into account all the parameters and variables present in a real industrial network. In a hybrid wired/wireless system, this problem is further complicated by the presence of communication systems with very different features. On the other hand, the penetration of wireless communication in industrial applications is contin- uously increasing, and its impact on the overall communication system must be understood and carefully evaluated. For these reasons, students need hands-on approaches in addition to traditional teaching methods through lectures and simulations. An effective solution is to use a real testbed that enables stu- dents to discover by themselves through practical experiments how network performances are influenced by different design choices, parameter settings, and operating conditions. The results obtained this way provide, from the student per- spective, a higher degree of “trustworthiness” than assessments made using only analytical or simulation-based tools. Students can directly verify the validity of the design solutions being in- vestigated, i.e., identifying their advantages and disadvantages. In addition, they can define and test new strategies to improve the system performance. Some network testbeds have been recently developed and implemented, and many researchers have used them to evaluate specific applications in specific areas. Two main application ar- eas have been exploited, i.e., wireless sensor networks (WSNs) and wireless mesh networks (WMNs). In the sensor network area, several testbed solutions have been implemented. One of the most popular is MoteLab [7], which is a set of software tools for the management of a sensor network deployed in a large laboratory. It provides an Ethernet back channel with a central server for scheduling, node repro- gramming, and data logging, and allows students to access the testbed through a web interface. Twist [8] is similar to MoteLab but allows greater interaction with students and the evaluation of both hierarchical and flat WSNs. Both testbeds allow the boundary between simulated components and hardware to be shifted in such a way that an application can run on a simulated radio model or in the real hardware present in the testbed. All these testbeds use a wired Ethernet for management purposes. To avoid the cabling costs that are associated with Ethernet- based management in a large deployment, the Eidgenössis- che Technische Hochschule Zurich testbed [9] uses an out of 0278-0046/$25.00 © 2007 IEEE