Nandakishore Kushalnagar

Exploiting heterogeneity in sensor networks

The presence of heterogeneous nodes (i.e., nodes with an enhanced energy capacity or communication capability) in a sensor network is known to increase network reliability and lifetime. However, questions of where how many, and what types of heterogeneous resources to deploy remain largely unexplored. We focus on energy and link heterogeneity in ad hoc sensor networks and consider resource-aware MAC and routing protocols to utilize those resources. Using analysis, simulation, and real testbed measurements, we evaluate the impact of number and placement of heterogeneous resources on performance in networks of different sizes and densities. While we prove that optimal deployment is very hard in general, we also show that only a modest number of reliable, long-range backhaul links and line-powered nodes are required to have a significant impact. Properly deployed, heterogeneity can triple the average delivery rate and provide a 5-fold increase in the lifetime (respectively) of a large batten-powered network of simple sensors.

Design and deployment of industrial sensor networks: experiences from a semiconductor plant and the north sea

Sensing technology is a cornerstone for many industrial applications. Manufacturing plants and engineering facilities, such as shipboard engine rooms, require sensors to ensure product quality and efficient and safe operation. We focus on one representative application, preventative equipment maintenance, in which vibration signatures are gathered to predict equipment failure. Based on application requirements and site surveys, we develop a general architecture for this class of industrial applications. This architecture meets the applications data fidelity needs through careful state preservation and over-sampling. We describe the impact of implementing the architecture on two sensing platforms with differing processor and communication capabilities. We present a systematic performance comparison between these platforms in the context of the application. We also describe our experience and lessons learned in two settings: in a semiconductor fabrication plant and onboard an oil tanker in the North Sea. Finally, we establish design guidelines for an ideal platform and architecture for industrial applications. This paper includes several unique contributions: a study of the impact of platform on architecture, a comparison of two deployments in the same application class, and a demonstration of application return on investment.

A stream-oriented power management protocol for low duty cycle sensor network applications

Most power management protocols are packet-based and optimized for applications with mostly asynchronous (i.e. unexpected) traffic. We present AppSleep, a stream-oriented power management protocol for latency tolerant sensor network applications. For this class of applications, AppSleep demonstrates an over 3/spl times/ lifetime gain over B-MAC and SMAC. AppSleep leverages application characteristics in order to take advantage of periods of high latency tolerance to put the network to sleep for extended periods of time, while still facilitating low latency responses when required. AppSleep also gives applications the flexibility to efficiently and effectively trade latency for energy when desired, and enables energy efficient multi-fragment unicast communication by only keeping the active route awake. We also present Adaptive AppSleep, an application driven addition to AppSleep which supports varying latency requirements while still maximizing energy efficiency. Our evaluation demonstrates that for an overlooked class of applications, stream-oriented power management protocols such as AppSleep outperform packet-based protocols such as B-MAC and S-MAC.

Intel Mote 2: an advanced platform for demanding sensor network applications

Intel Mote 2 is based on the PXA271 microprocessor featuring the Intel XScale® core. This multi-die package includes a high performance processor, 32 MB of flash and 32 MB of SDRAM. The processor integrates a DSP coprocessor, a security coprocessor and an expanded set of I/O interfaces (including UART, I2C, SPI, USB, CIF, I2S, and AC97) to ease sensor integration. The platform also provides an on-board 802.15.4 radio and the option to add other wireless interfaces via an SDIO interface. We have ported the TinyOS and Linux operating systems to this platform.

Sensor networks in Intel fabrication plants

The deployment of large-scale sensor networks in industrial environments presents technical challenges in achieving ease of deployment, flexibility in operation, and overall commercial viability. Sensor deployments are characterized by non-uniform placement of nodes, intermittent node connectivity, and the aggregation and reliable transfer of a large amount of data as networks scale to larger and larger sizes. Operation challenges include efficiently utilizing battery powered nodes, dynamically selecting sample periods and equipment clusters of interest, integrating with existing sensing and analysis infrastructure and easily correlating faults identified by the sensor network back to key factory operations and equipment. Commercial aspects involve returning value to the organization with hardened network nodes and reliable network operation that easily justifies the sensor network deployment and operating costs. We will demonstrate a sensor network in the ultra-pure water facility of an Intel fabrication plant, including details on the fab vibration application, the hardware nodes, and the heterogeneous wireless network. The fab vibration application uses vibration analysis to predict and correct equipment failures before fab operations are impacted. The ultra-pure water facility has a diversity of pumps and metal infrastructure both in a gymnasium size room and in an outside open air area that require 51 nodes with 201 vibration sensors. Data samples of 6K bytes are taken periodically from each vibration sensor, forwarded to a central server, measured for velocity, spike energy, acceleration, and displacement, analyzed with FFTs and trend analysis, and compared against expected profiles. When the analysis detects vibration variations outside of normal operating parameters, the affected equipment is scheduled for preventive maintenance, and repairs are made during normal down times. The hardware nodes for the fab vibration application include battery-powered motes (Figure 1(a)) and line-powered gateways. The motes are powered by four C Cell batteries, use a 900 MHz transceiver, and support an RPM (Revolutions per Minute) sensor and up to six vibration sensors. The gateways contain both a 900 MHz transceiver and a 2.4 GHz transceiver running 802.11b to bridge the sensor radios to the 802.11 network. Both motes and access points use heavy metal housings and are hardened for industrial use. Motes and gateways self configure into a heterogeneous wireless network on power up. Motes form an ad hoc network and cluster to gateways. Gateways form an ad hoc 802.11b overlay network and provide high-capacity data transport allowing us to scale the network. One of the gateways is connected to the corporate network through Ethernet. The motes are cycled through planned sleep and wake periods to alternately conserve power and perform vibration data acquisitions. Motes send the data to the gateways where it is aggregated and ultimately forwarded to a server for analysis and storage. Our demonstration includes the display of actual vibration data with real time updates from an Intel fab, together with hands-on samples of gateways, motes, vibration sensors, and RPM sensors used in the deployment. Overview of the network topology and physical deployment is also shown along with the key network technologies developed and used to run the vibration application. We also discuss current learnings from the deployment and future work.

Heavy industry applications of sensornets

Sensors are a cornerstone of heavy industrial operations. Manufacturing plants and general engineering facilities, such as power plants or shipboard engine rooms, require a high degree of sensing to ensure product quality and/or efficient and safe operation. Wireless sensor networks are a natural fit to meeting the demands of scale, data access and cost. However, the nature and environment of industrial applications presents unique requirements for sensornets. To better understand these challenges, we conducted deployments in two settings: a semiconductor fabrication plant and an oil tanker. The context for the deployments was a condition based maintenance application which monitors machinery vibration to detect/preempt failures.