Final Demonstration
From NESTFE Wiki
Networked Embedded Systems Technology (NEST)
Basic Research Final Demonstration Plan
Contents |
Demonstration Overview
On Aug 29, the day of the "NEST Final Demonstration," each study will give a presentation of its results research and analysis. There will also be unifying demonstrations such as "hide and seek with PDA's" and "maintaining a secure area". The studies we currently have planned include but are not limited to those below:
Concept Of Operations
[This section should contain a detailed description on the demonstration concept of operations (CONOPS), including the operational scenario(s) being demonstrated, the hardware, software, and infrastructure capabilities, the interactions between system components and personnel, and the basic research topics it will demonstrate.]
The operational scenario for this basic research is maintaining the integrity of a previously secured outdoor area, although the technology applies to many other scenarios. This task involves efficiently establishing a dense sensor array covering the area that can be maintained for long periods, both from a sensor lifetime and an overall network health and management viewpoint. It must detect the presence and track the movements of personnel and vehicles and report this information in a reliable, timely, and useful manner to a variety of fixed or mobile observational assets, including individuals that might be patrolling region or actively responding to notification, as well as automated assets that are capable of high fidelity inspection of identified areas. It should be able to detect and report various suspicious actions that deviate from expected behavior. It must also be able to operate in various degraded states while reporting the health and coverage of the network. Its operational modes must be evolvable over time through robust reprogramming of the network.
Each lowest tier network node contains a Telos Rev B module, an XSM sensor board, an adapter board between the two, a solar rechargeable power unit, and an outdoor enclosure. All sensor network communication is encrypted. The emphasis of this plan is not to develop new sensing hardware or modalities, so only the sensors from the XSM are used: PIR, acoustic, and magnetic.
In the scenario, the network observes a grassy field that contains some trees and a road, classifying and reacting to aberrant environmental behavior such as foot traffic leaving the road. Friendly personnel may move in the environment while carrying a PDA that provides tracking information and network status. The PDA also allows feedback and control of the network for sensor calibration and management of misbehaving nodes.
This commonly occurring scenario forms the basis for many more complex scenarios and drives the basic research, investigation, and understanding of core capabilities of large-scale (multi-hundred to multi-thousand node) wireless sensor network deployments, including the following:
Perpetual deployment
We will demonstrate and analyze an advanced wireless node that extends the series of NEST Open Experimental Platforms. Its sensors and mechanical design are based on the XSM node, utilizing the XSM board for sensors and the enclosure concept. The XSM will be adapted to the newly designed Telos system board, which provides improved power characteristics, storage, reliability, and an IEEE standard radio. Intelligent power harvesting from solar with careful management of recharge cycles will provide lifetime limited only be the physical integrity of the node.
Security
All data traffic will be encrypted using hardware support provided on the Telos node. Key management strategies for large-scale deployments will be evaluated. In addition to the key management and encryption, we propose the following scenario for breaching the security and the countermeasure for it. We assume that some of the nodes in the network are fully compromised. The attacker will then use these nodes to alter the tracking information in order to fool the network about the location of the intruder. As a countermeasure, we propose to build a trust map of the sensor network at the base station,and update it periodically. In addition, each node builds a neighbor table and ranks its neighbors based on reputation and trust relation. The combination of the base station and node trust maps will help in isolating the compromised nodes,and reliable data aggregation.
Large-scale management
A low-overhead, low-power, flexible network management facility will be demonstrated and evaluated. It must be easy to augment system and application code to enable management. The network operator must be able to easily express and obtain attributes and regions of interest. The protocols should scale with the level of management activity, dropping essentially to zero energy usage when management is inactive, operate even when other network layers or services are faulty, and have a very limited footprint.
Self-localization
The many individual nodes should be able to determine their absolute or relative positions with little manual intervention, and in a robust and stealthy manner. We will demonstrate the effectiveness of RF-based techniques utilizing capabilities of the new radio starting from a limited number of known anchor nodes as well as the use of mobile assets, such as people, UAVs, or UGVs, with other localization facilities, such as GPS, to interactively reduce localization uncertainty.
Robust programming and Rapid Retasking
It must be possible to reliably deliver complete binary images of system or application code to large or focused subsets of nodes and to maintain consistency across the set even when nodes are added, die, or are intermittently connected. We will demonstrate and evaluate the deluge family of robust dissemination algorithms and extensions to ensure integrity. It should also be possible to rapidly retask the network among a family of possible behaviors. In addition to reparameterization, we will demonstrate the use of application specific virtual machines built upon the Mate’ framework to allow a rich retasking through dissemination of protected bytecodes.
Network communication
The system capabilities above, as well as well as the application capabilities below fundamentally rest upon four basic network communication capabilities: (1) robust dissemination of information to a large collection of nodes, (2) reliable collection of information, (3) efficient exchange of information among physically localized groups of nodes, and (4) routing of information from any (potentially mobile) point to any (potentially mobile) point. Each class of network communication will be demonstrated and evaluated in isolation and in the context of various operational settings.
Sensing and Identification
Individual nodes have the capability to perform local sensing and signal processing. We will demonstrate significant steps toward passive vigilance, where the energy expended in sensing is proportional to detections, rather than time spent observing. The key concept is the sensor cascade, in which low-power, low-fidelity sensors with hardware wake-up capabilities can invoke selectively higher level, higher capacity assets. In addition, collections of nodes share processed information to refine the detection and classification.
Visualization
It must be possible for human operators to observe the features which the network has detected, as well as the health and status of the network itself. In both cases, the operator needs to be apprised of the certainty or uncertainty of the findings in order to plan responses. This information needs to be provided not only to fixed assets monitoring the network, but to handheld mobile assets moving through the network itself.
Tracking
Identified objects should be tracked and estimated trajectories reported to a variety of assets. We will demonstrate a range of difficulties, including reporting to one fixed asset, reporting to many fixed assets, and reporting to one/many mobile units.
Asset coordination
In addition to reporting real time network and Ops information to fixed and mobile human assets, we seek to demonstrate a level of automated response. This potentially takes two forms. In response to a detection, a remote controlled aerial vehicle with camera mount will be directed to point of activity to gain further fidelity or autonomous unmanned ground vehicles will pursue the detected object. Such asset coordination introduces a closed loop feedback that will be utilized in several other aspects of maintaining the system, including improving localization accuracy by actively moving to and obtaining the position of nodes with high uncertainty, as well as repairing various network faults.
Objectives
[This section should contain information on the specific objectives of your demonstration, both operational and technical.]
The primary objective is to study and characterize for a thousand node network each of the technical studies listed in Section 1.1. Criteria for each study are listed in Section 4. The operational objective is to deploy and maintain a thousand node perpetual network and to track and classify foot traffic. The operational objective is a formulation and showcase for the technical studies.
Execution Activities
[This section should contain a step-by-step description of the activities required to set up, execute, and debrief/evaluate your demonstration.]
The network is deployed weeks before the official demonstration. Then, on the day of the demonstration
- The network is initialized
- The sensor network is woken up
- The sensor network forms a robust communication network
- Network and node health is assessed, showing bulk node data collection
- Reprogramming and/or re-tasking of the network may occur, showing bulk data dissemination
- The network performs self-localization
- Additional features of the network are shown
- Any-to-any network communication
- Calibration and additional localization through PDA's and/or aircraft
- The network enters its tracking modality and is given things to track
- Reporting to one/many fixed assets
- Reporting to one/many mobile assets (PDA's)
- The tracking alarms are responded to
- By people with PDA's
- By aircraft with camera or ground robots
- Debrief by discussing the measurements and performance taken for each component described by the technical studies
Demonstration Requirements
[The following sections should contain detailed information on the components required for your demonstration. Please be as specific as possible and, where applicable, provide performance, power, timing, memory, etc. specifications. For NEST Middleware Services, please list each service separately and provide as much detail as possible on the algorithms, interfaces, data structures, etc. Please use the Other section to list demonstration requirements that don’t fall into one of the other categories.]
Hardware
- 1300 XSM boards delivered by Program Manager to UCB for the Trio's, below
- 1300 Trio sensor nodes manufactured by UCB, each consisting of the following components:
- Telos Rev B (MSP430 F1611 with 10k RAM and 48k ROM, 1 Mbyte code flash, CC2420 802.15.4 radio at 2.4 GHz), using onboard antenna
- Prometheus solar recharging power system
- XSM board providing PIR, acoustic, and magnetic sensing
- Interface board transforming between the Telos Rev B 16-pin connector and the XSM 51-pin connector
- Weather resistant enclosure designed to fit these components (dimensions: 4in x 4in x 4in)
- [300 of these Trio nodes will be delivered to Program Manager for use by other contractors]
- ~3 large, high gain 2.4 GHz antennas to assist network management and data collection
- ~20 Stargate-class Tier 2 devices
- 10 PDA's as mobile stations (likely units: Sharp Zaurus)
- 5 to 10 mobile GPS units, preferably attached to the PDA's
- 2 to 3 PC fixed base stations with plasma or projection displays
- 2 remote-controlled aerial fixed-wing aircraft with mounted cameras
- ~5 autonomous UGV robots, each with a Telos for motor control, PDA for high level computation, and GPS for navigation
Software
- Trio sensor node embedded software
- Power aware operation
- 802.15.4 encryption management
- Large-scale management (SNMS: Sensor Network Management System)
- Localization: calibrated RF signal strength
- Network reprogramming (Deluge)
- Network communication
- Any-to-any routing: geographic or other
- Data dissemination (Drip, Deluge)
- Data collection (Drain)
- Sensing
- System wake-up on event
- Signal processing
- Rapid re-tasking with virtual machines (Maté)
- Tracking
- Multi-object tracking
- Behavior classification
- Report to ...
- one fixed asset
- many fixed assets
- report to mobile units (PDA's)
- Sensor calibration
- PDA and PC control and display software
- Display field, node state, detections, classifications, uncertainty, live camera feeds
- Reprogram network
- Re-task network
- Reconfigure network
- Control and calibrate nodes: sensing, localization, communication
- Data gathering and logging
- Pursuit advice for people and aircraft
- Highest level pursuit control for ground vehicles
- Unmanned ground vehicle
- Embedded software
- Control and management software
Nest Middleware Services
- Producing, providing, and using
- Deluge bulk data dissemination for network reprogramming
- SNMS network management
- Drip data dissemination (not bulk)
- Drain data collection
- Mobile-pairs any-to-any and few-to-few routing
- Neighborhood management and sharing
- Maté virtual machine for rapid re-tasking
- TOSBoot and DefaultImage reboot and failsafe support
- Time synchronization (Vanderbilt)
- Geographic Mesh routing (UT)
Connectivity/User Interface
Connectivity and user interface are provided through PC base stations displayed for observers on plasma screens or projectors – necessary for possible outdoor display in bright sunlight. Each PC is connected to a Telos using a high-gain antenna. All PC’s log all data to a local or centralized database for later analysis and retrieval.
PDA connectivity is achieved by connecting a Telos to the PDA serial port. Its user interface is based as much as possible on the PC interface, implying both interfaces are written in a cross-platform manner, likely in Java. Some or all PDA’s have a mobile GPS unit to assist localization and calibration.
Personnel
Demonstration will be performed by UCB personnel.
Other
As noted in hardware section, the sensor boards will be XSM boards provided by the Program Manager. Arrival of these boards is time critical in the deployment plan.
The final demonstration will cap off the longer term use. It is to have been operational in the field for two months prior to that demonstration. Extensive data collection, analysis, refinement of distributed algorithms and mechanism at scale to take place over the longer term leading up to the final.
Demonstration Architecture
[This section should contain a diagram (or diagrams) of your demonstration system architecture that contains all system components and illustrates the interfaces between components. This section should also contain a description of the overall architecture and detailed information on the data flows between components, message timing, and protocols.]
Much of the demonstration architecture is active research with conference publications that more completely and precisely describe the interfaces, data flows, message timings, and protocols. The following papers provide detail on key elements.
- BMAC-sensys04.pdf
- Low power communication
- Description: A low-power carrier-sense media access protocol for energy efficient sensor network applications.
- Deluge-sensys04.pdf
- Reliable network reprogramming
- Description: A reliable bulk data dissemination protocol used for network reprogramming.
- Hood-mobisys04.pdf
- Neighborhood communication
- Description: A neighborhood programming abstraction that fits well with many existing sensor network applications.
- LocRSSI-ucb04.pdf
- Radio signal strength localization
- Description: A study and experience paper describing how to get ad-hoc signal strength localization to work in practical environments.
- LocDeploy-ucb04.pdf
- Localization experiences
- Description: Empirical analysis of localization in a sensor field using ultra-sonic ranging.
- Mate-osdi04.pdf
- Virtual machines for retasking
- Description: Application specific virtual machines
- PEG-ewsn05.pdf
- 100 node deployment experience
- Description: Design and implementation of a pursuer evader game - from the NEST demo in the summer of 2003.
- Prometheus-spots05.pdf
- Solar rechargeable power system
- Description: A two stage storage system consisting of super capacitors (primary buffer) and a lithium rechargeable battery (secondary buffer).
- SNMS-ewsn05.pdf
- Sensor network management system
- Description: designed to be simple and have minimal impact on memory and network traffic, while remaining open and flexible.
- Telos-spots05.pdf
- Ultra-low power sensor network module
- Description: the latest in a line of motes developed by UC Berkeley to enable wireless sensor network (WSN) research.
Demonstration Evaluation
[This section should contain a detailed set of evaluation criteria and performance metrics that can be used to assess how well the demonstration met the basic research objectives. Please be as concrete and specific as possible and quantify all evaluation criteria and/or performance metric values.]
- Perpetual deployment: energy consumption and generation. Achieved duty cycle. Node loss rate.
- Security: cost of encryption, key mgmt overhead.
- Large-scale management: Time to perform mgm command vs count/extent, Time to collect mgmt data vs count/extent, Mgmt bandwidth/overhead, Scaling limits. Mgmt effectiveness with active load.
- Self-localization: Time to localize, position accuracy, number of anchor nodes, response to active improvement
- Robust programming and Rapid Retasking:
- Time, message complexity, redundancy, loss rate in reprogramming vs node count/extent and image size.
- Time, message complexity, redundancy, loss rate in parameter changes vs node count/extent
- Time, message complexity, redundancy, loss rate in script dissemination vs node count/extent. Script expressiveness. Script safety.
- Consistency maintenance traffic for each of the above.
- Time & cost of bring joiner node to consistency.
- Network communication:
- Achieved bandwidth and latency vs extent & workload
- Rate of adaptation in response to mobility / node failure / interference conditions
- Reliability & retransmission cost
- Capacity utilization
- Stretch factor
- Energy usage
- Sensing and Identification:
- Detection rate, detection delay, false positive rate, identification accuracy
- Selectivity at each level of sensor cascade
- Sensing energy per unit coverage area.
- Notification delay and rate
- Visualization: quality and scale of information presentation.
- Tracking: uncertainty, object trajectory limits, disambiguation rate
- Asset coordination:
- Fraction of events 'captured', time to capture
- Degree of uncertainty improvement.
- degradation with respect to position noise and estimate delay
Demonstration Schedule
[This section should contain a workplan (preferably a Microsoft Project Gantt chart) that details the activities associated with your demonstration from initial concept and requirements development through demonstration execution. Please include start/end dates for each task, any task dependencies, and key milestone and deliverable dates.]
- Design and manufacture of 1,300 Telos Boards
- Design complete 2004, 1,000 boards manufactured and tested 1/19/2004. Remainder to be built and tested in Feb.
- Sensor Board.
- Design and manufacture complete by OSU/XBOW prior to PI meeting
- Received 100 which will be used in enclosure design and test.
- Feb 10. Receive from Program Mgr 1,300 boards for deployment.
- Trio Design.
- Enclosure, adapter,
- Prometheus power system (done)
- Start: 11/04, End: Feb 10
- Depends: ranging studies in the field with Telos RevB on XSM
- Tier 1 Node build and Assembly
- Initial Set: March 1
- Complete Set: March 21
- Depends: Arrival of XSM boards by Feb 15.
- Design, acquire, deploy Tier 2
- Purchase high-gain antennas, stargate-class devices
- Initial portion: Apr 1, Final May 1
- Deploy Tier 1
- Test set: 100 nodes - Apr 1
- Scaling: 400 nodes - May 1
- Final: 1,000 nodes - June 1
- Visualization
- Large operations display.
- SW Spec: March 1, Prototype: Apr 1. Rev 1: June 1, Final Rev: Jul 15.
- Mobile PDAs.
- SW Spec: March 1, Prototype: Apr 1. Rev 1: June 1, Final Rev: Jul 15.
- Large operations display.
- Services
- General service schedule (unless otherwise noted)
- Must be well-defined with a SW spec by March 1
- Prototypes by Apr 1
- Working demos May 1
- Scaling analysis June 1
- Power aware operation
- 802.15.4 encryption management
- SNMS
- Localization: calibrated RF signal strength
- Deluge
- Network communication: SW Spec: March 1
- Any-to-any routing: geographic or other
- Data dissemination (Drip, Deluge)
- Data collection (Drain)
- Sensing
- System wake-up on event
- Signal processing
- Rapid re-tasking with virtual machines (Maté)
- Tracking
- Multi-object tracking
- Behavior classification
- Report to ...
- one fixed asset
- many fixed assets
- report to mobile units (PDA's)
- Sensor calibration
- PDA and PC control and display software
- Display field, node state, detections, classifications, uncertainty, live camera feeds
- Reprogram network
- Re-task network
- Reconfigure network
- Control and calibrate nodes: sensing, localization, communication
- Data gathering and logging
- Pursuit advice for people and aircraft
- Highest level pursuit control for ground vehicles
- Unmanned ground vehicle
- Embedded software
- Control and management software
- General service schedule (unless otherwise noted)
- Final demonstration. July 31
- Reports and post demonstration evaluation.
