Speaker: David E. Culler, Arched Rock Corporation & University of California, Berkeley
Title: Wireless Sensor Networks as the Next Tier of the Internet

Abstract:
Instrumentation and computing have co-evolved for fifty years with improvements in circuit, storage, interconnection, and software technologies. However, computing has experienced explosive growth through universal connectivity, culminating in the web, while instrumentation has largely been constrained by proprietary, domain-specific networks. The wireless sensor networks emerging from the thousands of companies and universities developing technology and applications on open platforms, such as motes and TinyOS, are likely to change both of these worlds dramatically - especially as sensornets converge with internet technologies. No longer will the web be limited to human keystrokes, clicks and swipes, it will be infused with rich physical information sources. Instruments and sensors will feed directly into the enterprise creating actionable awareness of business processes. This talk will describe core challenges, opportunities, and advances in several levels of technology that serve to make wireless sensor networks the next tier of the internet.

Speaker: Professor Mani Srivastava, UCLA
Title: Energy-aware Wireless Sensor and Actuator Networks

Abstract:
Ensuring adequate system lifetime while sustaining adequate quality of service continues to be a significant bottleneck for many sensor network deployments. Proper addressing of this problem requires going beyond low-power chip and nodes to energy-aware operation of entire fields of networked nodes. Crucial is the energy impact of the interplay between the computation, wireless communication, sensing, and actuation aspects of these systems.

After a discussion on the factors that effect sensor network lifetime, and an overview of various techniques for energy management, the talk will focus on several emerging techniques for energy-aware operation. First, the talk will look at the energy problem in ultra low duty cycle application, and show how emerging predictive and rate adaptive time synchronization mechanisms can yield one to two orders of magnitude reduction in radio energy consumption relative to conventional radio power reduction mechanisms. Next, the talk will explore the emerging sensor platforms equipped with environmental energy harvesting, and describe harvesting-aware system resource management techniques that yield significant improvements in system lifetime and promise "energy neutral" operation. Lastly, the talk will describe how, rather counter-intuitively, actuation has emerged as a key to the lifetime problem, particularly in addressing energy issues surrounding the sensors themselves, which although usually ignored are often the energy bottleneck.

Speaker: Dr Kay Roemer, ETH Zurich
Title: Programming Models for Sensor Networks

Abstract:
Today, implementing and deploying a sensor network application is a tedious task whose successful completion typically requires significant experience and expertise with the underlying sensor node hardware and operating system. This is in stark contrast to the anticipated use of large-scale sensor networks for real-world applications, where application-domain experts -- rather than system experts -- should be able to customize and deploy sensor networks.

In this talk, we argue that novel programming models are needed to raise the level of abstraction from system-centric programming to application-centric programming of sensor networks. In particular, such models can support programming a sensor network as a whole (rather than individual nodes) using declarative specifications of the functionality that needs to be realized. The provision of such programming abstractions is, however, a challenging task as it is hard to combine expressiveness and simplicity of the model with resource-efficient execution on sensor nodes.

We illustrate the concept and involved challenges by presenting an example from our recent work that is called "generic role assignment." This programming model supports the assignment of user-defined roles to sensor nodes such that certain conditions are met. Many common network configuration problems such as coverage (assign roles ON and OFF to sensor nodes such that ON nodes cover a physical area with their sensors), clustering, or in-network data aggregation can be formulated as role assignment problems that can be supported with our programming model.

Speaker: Professor David Whalley, Florida State University
Title: Improving Program Efficiency by Packing Instructions into Registers

Abstract:
This presentation is about a novel architectural and compiler approach to simultaneously reduce energy consumption and decrease code size by having the compiler place the most frequently occurring instructions into an instruction register file (IRF). Multiple entries in the IRF can be referenced by a single packed instruction from ROM or the L1 instruction cache. A number of enhancements are used to increase the packing level, which include parameterization of immediate values and branch displacements, use of register windows, and a variety of compiler optimizations such as instruction selection, register renaming, and scheduling. In addition, the IRF is shown to complement other architectural features to save power, such as L0 and loop caches. The experimental results show significant improvements in energy consumption and code size, as well as small improvements in execution time.

Speaker: Professor Sanjay Jha, UNSW
Title: Designing a Hybrid Sensor Network

Abstract:
In the past couple of years, the development of embedded system technologies have made it feasible to deploy large-scale sensor and actuator networks. These networks are revolutionizing the way in which we understand, monitor and control complex physical environment, and provide one of the missing connections between the Internet and the physical world. Because of size, form factor and cost considerations, wireless sensor networks suffer from severe resource constraints, such as communication bandwidth and range, computation power, memory and energy. Furthermore, sensor networks are expected to consist of hundreds of sensors in the future.

This talk will address the challenges involved in designing robust and scalable sensor network system. In contrast to the flat networking model considered in previous sensor networks research, we have proposed a hierarchical or hybrid network architecture which is more scalable and robust. Hybrid sensor networks consist of resource-impoverished sensors and resource-rich sensors, called micro-servers. Because of the different capabilities of heterogeneous devices in hybrid sensor networks, there is need for effective deployment and utilization of network resources. Therefore, we introduce resource provisioning and management algorithms to optimize the performance of hybrid sensor networks.

A native Australian sensor network application, cane toad monitoring, motivates this research on resource provisioning and management of hybrid sensor and actuator networks. This is a challenging sensor network application characterized by high frequency sampling, complex signal processing for in-network reasoning, and wide-area sensing coverage. This talk will discuss design and implementations of a hybrid sensor network system tailored to address these challenges as well as the design and evaluation of a reverse tree-based Anycast communication paradigm for hybrid sensor and actuator networks.

Speaker: Dr Cristina Cifuentes, Sun Microsystems
Title: Java(TM) on Wireless Sensor Devices

Abstract:
In the last few years we have seen a huge increase in the "buzz" around the topic of Wireless Sensor Applications. Products such as Motes and Embers have emerged, along with standards including IEEE 802.15.4 and Zigbee. We have seen reports of a few applications in journals and at conferences, but on the whole the developer community has not moved significantly to create applications for these new platforms.

The Sun(TM) Small Programmable Object Technology (SPOT) project at Sun Labs was started to draw together several threads of complementary research into a new advanced development programme. The result of this program is a new system designed to allow developers to rapidly create complete Wireless Sensor Applications using Java. The system includes hardware--small battery powered sensor devices; and software--a tiny reflective Java VM (the Squawk VM) and a development kit. Squawk acts as the OS all the way down to the hardware, not just as a thin veneer
over another system.

In this talk I will introduce the Sun SPOT technology and the Squawk Java virtual machine. Squawk is a Java ME compliant VM written in Java that implements an application isolation mechanism. Isolates represent applications as objects, thereby allowing for the reification of applications. Using Squawk, applications can be migrated from between Sun SPOTs. The Squawk wireless API allows developers to write wireless sensor applications using radio and radiogram connections. This presentation will include demonstrations of the Sun SPOT and Squawk technologies.

Presenters: Professor Subhash Challa, UTS and Associate Professor M Palaniswami, University of Melbourne
Title: Innovative Applications of Sensor Networks in the Australian Context

Abstract:
Breakthrough technologies in sensing, information and communications technologies and the associated technological convergence including socio-economic imperatives are driving the sensor network revolution. This presentation is focussed on innovative applications of these ground breaking technologies in the Australian context. The applications include Heritage Listed Remote Wodden Bridge Monitoring, Electricity Load Forecasting, Optimization of Acquaculture Yeilds, In-stream waste water management, Smart Surveillance, Virtual Land Mines, Great Barrier Reef Monitoring and others. Several companies like ADI, Road Transport Authority, Rolachem, Cradle Technologies, Iomniscient, Intersystems, Defense Science and Technology Organization, Reline Technologies with stakes in the emerging market opportunities in this space are supporting innovations in this space in Australia. The strategic positioning of these companies and the role these technologies play in building cost-effective solutions in this space will also be reviewed in this presentation.

Speaker: Dr Pavan Sikka, CSIRO ICT Centre, Brisbane
Title: Wireless Ad hoc Sensor and Actuator Networks on the Farm

Abstract:
Agriculture accounts for a big portion of the GDP in most developed countries. However, farm management, particularly of large-scale extensive farming systems, is hindered by lack of data and increasing shortage of labour. We have deployed a large heterogeneous sensor network on a working farm to explore sensor network applications that can address some of the issues identified above. Our network consists of a variety of sensor nodes, both static and mobile, that provide information on the state of the farm and the animals on the farm. The whole network is connected to the world at large via a high-gain radio link to a gateway on the farm which in turn is connected to our network over an ISDN line. All static nodes, including the radio link, are solar powered.

The initial goals of the deployment are to provide a testbed for sensor network research in programability and data handling while also being a useful tool for scientists to study animal behavior. In the longer term, we aim to use the deployment for exploring farm management systems that could transform the way farms are managed.

Speaker: Dr Kerry Taylor, CSIRO ICT Centre, Canberra
Title: RFID for Aircraft maintenance

Abstract:
RFID technology is the basis for the largest sensor networks today, and is certainly a rapidly growing technology in terms of both the sizes of networks and the applications to which they are being directed. I will talk about some of these applications, but then will focus on the intention to tag major parts in the new 787 Dreamliner to improve aircraft maintenance procedures. There are important implications for data management. Firstly there are new opportunities to make use of use the capacity for deeper and more immediate visibility of the maintenance operations, which stresses conventional information architectures. Secondly, there is a need to manage the structural and functional relationships between aircraft parts in a computationally scalable manner, so that tagged parts may be connected into the wide range of information resources focused around a part. I will talk about our current work addressing both of these problems, based on the development of an ontology-enabled data and service integration system on one hand, and the development of large real-world, scalable and reasonable ontologies on the other.

Images used with the permission of Crossbow Technology Inc http://www.xbow.com/index.aspx

Page updated: 16-Feb-2006