Sally K. Wahba

Teaching CS unplugged in the high school (with limited success)

CS Unplugged is a set of active learning activities designed to introduce fundamental computer science principles without the use of computers. The program has gained significant momentum in recent years, with proponents citing deep engagement and enjoyment benefits. With these benefits in mind, we initiated a one-year outreach program involving a local high school, using the CS Unplugged program as the foundation. To our disappointment, the results were at odds with our enthusiasm --- significantly. In this paper, we describe our approach to adapting the CS Unplugged materials for use at the high school level, present our experiences teaching it, and summarize the results of our evaluation.

An Empirical Evaluation of Embedded Link Quality

We present the results of extensive field experiments designed to evaluate the precise relationship between wireless link quality, radio power level, and transmission distance in outdoor embedded deployments. We focus on ZigBee networks, using Tmote Sky nodes, a popular commercial hardware platform for wireless sensor systems. The results are used to develop a piece-wise regression model suitable for pre-deployment link quality prediction. We characterize the contexts in which the model provides high accuracy, and discuss potential threats to validity.

A Testbed for Visualizing Sensornet Behavior

We present a testbed for visualizing the behavior of sensor network applications constructed using TinyOS 2.0. The testbed enables network designers to capture node-level runtime behaviors and to link the captured behaviors to form a network-level snapshot. A multi-resolution approach based on static selectors and dynamic filters helps to achieve manageable visualizations by including only those network nodes and program actions relevant to a global behavior of interest. We demonstrate the systems utility using a standard network example and summarize its resource usage and performance characteristics.

Initiating a design pattern catalog for embedded network systems

In the domain of desktop software, design patterns have had a profound impact; they are applied ubiquitously across a broad range of applications. Patterns serve both to promulgate expert knowledge and as a vocabulary for documenting software design. The result is higher-quality software, reduced development effort, and improved documentation. This impact has yet to be felt in the emerging domain of embedded network software. The applications are characterized by highly distributed, reactive computations executed over resource-constrained processors; they present a host of new design challenges. Without a coherent repository of expert solutions, best practices do not emerge, design flaws are repeated, and developer communication is hindered. In this paper, we initiate a pattern catalog for embedded network software. The contributions are two-fold. First, we present three new design patterns distilled from a careful analysis of expert-crafted software. Second, we present a formalism for specifying the structural requirements of such patterns. The resulting specifications provide a rigorous foundation for pattern identification and documentation.

Fast distributed simulation of sensor networks using optimistic synchronization

Network simulation is an important tool for testing and evaluating wireless sensor network applications. Parallel simulation strategies improve the scalability of these tools. However, achieving high performance depends on reducing the synchronization overhead among simulation processes. In this paper we present an optimistic simulation algorithm with support for backtracking and re-execution. The algorithm reduces the number of synchronization cycles to the number of transmissions in the network under test. We implement SnapSim, an extension to the popular Avrora simulator, based on this algorithm. The experimental results show that our prototype system improves the performance of Avrora by 2 to 10 times for typical network-centric sensor network applications, and up to three orders of magnitude for applications that use the radio infrequently.

A metadata encoding for memory-constrained devices

With the broad applicability of wireless sensor networks across fields, it is desirable to develop self-describing sensor nodes that can operate in a plug-n-play manner. In this paper, we present MoteML, a metadata encoding suitable for storage on memory-constrained devices, designed in support of this goal. MoteML is consistent with Sensor Web Enablements [23] Sensor Model Language (SensorML). More specifically, while MoteML does not conform to the SensorML schema, it can be translated into SensorML and vice-versa. This paper explores the available solutions for storing self-describing information on memory-constrained sensor nodes and presents the design of MoteML. MoteML is a text-based encoding that captures a subset of SensorML in a template-based structure. This text data is then compressed using available text compression techniques. The resulting file is small enough to be stored on a memory-constrained embedded device.

Visualizing the runtime behavior of embedded network systems: A toolkit for TinyOS

TinyOS has proven to be an eective platform for developing lightweight embedded network applications. But the platform’s lean programming model and powerecient operation come at a price: TinyOS applications are notoriously dicult to construct, debug, and maintain. The development diculties stem largely from a programming model founded on events and deferred execution. In short, the model introduces non-determinism in the execution ordering of primitive actions | an issue exacerbated by the fact that embedded network systems are inherently parallel, distributed, and reactive. The resulting set of possible execution sequences for a given system is typically large and can swamp developers’ unaided intellectual ability to reason about program behavior. In this paper, we present a platform-neutral visualization toolkit for TinyOS 2.0 to aid in program comprehension. The goal is to assist developers in reasoning about the computation forest underlying a system under test and the particular branches chosen during each run. The toolkit supports comprehension activities involving both local and distributed runtime behavior. The constituent components include (i) a full-featured static analysis and instrumentation library, (ii) a selection-based probe insertion system, (iii) a lightweight event recording service, (iv) a trace extraction and reconstruction tool, and (v) three visualization front-ends. We demonstrate the utility of the toolkit using both standard and custom system examples and present an analysis of the toolkit’s resource usage and performance characteristics.

Fast Distributed Simulation of Sensor Networks Using Optimistic Synchronization

Network simulation is an important tool for testing and evaluating wireless sensor network applications. Parallel simulation strategies improve the scalability of these tools. However, achieving high performance depends on reducing the synchronization overhead among simulation processes. In this paper, we present an optimistic simulation algorithm with support for backtracking and re-execution. The algorithm reduces the number of synchronization cycles to the number of transmissions in the network under test. We implement SnapSim, an extension to the popular Avrora simulator, based on this algorithm. The experimental results show that our prototype system improves the performance of Avrora by 2 to 10 times for typical network-centric sensor network applications, and up to three orders of magnitude for applications that use the radio infrequently. We also implement a distributed version of SnapSim, D-SnapSim, which runs on a cluster. The experimental results show that D-SnapSim further improves the performance of SnapSim by up to 10 times for applications that use the radio frequently.

NePTune: Optimizing Sensor Networks

We present the NePTune approach and supporting system architecture for sensor network optimization. NePTune relies on a control loop strategy with performance monitoring, dynamic source code generation, and network reprogramming. We present an application of the system architecture and explore its efficacy in the context of a resource utilization problem - specifically, to minimize the memory consumption of a neighborhood management service. All experiments are conducted using a physical network testbed consisting of 80 Tmote Sky nodes.

An empirical analysis of communication links in embedded wireless networks

Embedded wireless networks have been widely used, from social networking to disaster response. Despite their increasing application, developers still face many challenges. One of the most significant is ensuring reliable communication between communicating devices. In this paper, we present an empirical examination of some of the key factors that affect the quality of wireless links in outdoor environments. Specifically, we study the effects of height differences between devices, device orientation, and data transmission rate on communication quality.