Final Report Summary - COMMONSENSE (Development of a Common Sensor Platform for the Detection of IED "Bomb Factories")
Executive Summary:
The project has achieved significant improvements over the state of the art, both in terms of detection limits and novel analysis algorithms. Furthermore, the integration of sensor units into modules (both standalone and wireless-interfaced units) and subsequent testing emphasises the exploitation potential of the COMMONSENSE platform and its potential benefit for EU citizens and industry.
Project Highlights
1. Nanoelectrochemical sensors and packaged modules were developed for solution-based detection and discrimination of military-grade nitroaromatic and nitroester explosives and related compounds (e.g. DNT, TNT, PETN) with limits of detection (LoD) below 10 ppb for TNT.
2. Fluorescence-based optical “turn-off” sensors and modules incorporating novel organic semiconductor materials prepared as gas-permeable membranes were developed. Detection limits down to ppb-level were achieved for gas-phase detection of TNT, DNT, NB, PETN and RDX for lab-based sensors, 100 ppb for integrated modules.
3. Integrated multi-sensor modules (with wireless monitoring and control) were also developed for final testing. Modules incorporated photoionisation detectors (LoD ~ 100 ppb for DNT, 78 ppb for TATP), electrochemical sensors (LoD ~200 ppb for TATP), fluorescence turn-off sensors and organic field-effect transistor sensors.
4. Door-mounted, small-area radiation sensor modules were developed and enabled wireless detection of 112 kBq sources (60Co, 137Cs) at a stand-off distance of 32 cm and a 30 kBq source (133Ba) at a distance of 16.5 cm. These modules demonstrated world-leading energy resolution and low false-alarm rate.
5. A secure, resilient and portable plug-n-play wireless communication network has been developed and validated. The network, based on IEEE 802.15.4 2006 standard and IETF standards (IPv6 & CoAP), is also compliant with US DoD JPEO-CBD Common CBRNE Sensor Interface (CCSI) and software gateway for CBRNE sensors. Applications interoperability was demonstrated and tested successfully.
6. An intelligent “learning” network was developed, using chemometric algorithms to teach the system to detect explosives and ignore interferents, yielding a “world-first” in discrimination of molecular isomers.
7. Customised/bespoke facilities were established for validation of COMMONSENSE sensors and modules developed for gas-phase detection of precursors, explosives and interferents at levels below 50 ppb.
8. An end-user testing site was established by PSNI with the aid of CAST (UK Home Office Centre for Applied Science and Technology).
A project website has been established at www.fp7projectcommonsense.eu to publicise the goals of the project, introduce the consortium and disseminate progress to the wider scientific community, industry and the public – details on peer-reviewed publications and presentations are available on the project web site.
Project Context and Objectives:
Objectives:
The detection of improvised explosive device (IED) manufacturing facilities is crucial for the security of citizens, as well as infrastructures and utilities. The Madrid (2004) and London (2005) bombings, as well as more recent events in Stockholm (2010) and Oslo (2011) have highlighted the need to tighten national security in the face of ongoing threats. The growing number of threats to the security of EC citizens may be traced by following the emergence of the nomenclature used to describe these threats. Where once the main threats were described as NBC (nuclear, biological and chemical), the current threats posed by terrorists, insurgents and criminals are labelled as CBRNE (Chemical, biological, radiological, nuclear and explosive).
Radiological weapons (the ‘R’ in CBRNE) lend themselves perfectly to spreading fear and long-term contamination of buildings or areas. The addition of ‘E’ for explosives is the latest extension of the acronym, and reflects the growing concern about the use of IEDs in urban environments. With the growing number of potential threats, it is no longer possible (or preferable) to use a single detector technology to ensure the security of citizens. Common sense dictates that if no single sensor technology exists to detect all types of explosive compounds, we must use the established technologies in combination to fully protect EC citizens. The need clearly exists for a single distributed network, with a common interface and communications protocol, to manage and communicate with a variety of different sensor technologies, and use the combined sensor data to produce clear and unequivocal results with low false positive/negative readings.
The purpose of the COMMONSENSE project is to provide greater protection to EU citizens from the threats posed by improvised explosive device (IED) “bomb factories” in urban environments without the use of intrusive surveillance and tracking techniques. COMMONSENSE has developed a very low power, portable, resilient and plug-n-play wireless sensor network for integrated detection of explosives, precursors and radio-active species in gas and water. The partners will produce a series of novel organic, polymeric and nanocrystalline materials with tuned optoelectronic properties and surface affinities to be used as the active sensor elements. These elements will be incorporated into devices based on optical, electrical, and other readout mechanisms, for detection of airborne and waterborne analytes.
The key point in the use of such a variety of sensor technologies is that no one substance can act as an interferent to all of the sensors, thus reducing false positives and negatives. Eliminating the remaining false readings will be achieved through use of the chemometric algorithms in order to teach itself to recognise the “fingerprint” sensor response to different explosives types and ignore interfering compounds. The COMMONSENSE network is intended to be deployed in European urban environments pervasive network, using low-cost sensor modules in wireless communication with a central server, which manages remote access, control, operation and readout from this network. Development and implementation of the COMMONSENSE will therefore enhance the daily security of European citizens via continuously monitoring of their home and work environments, and alerting security forces and other responders to the presence of IED manufacturing facilities.
The main objectives of the COMMONSENSE project are as follows:
• Development of modules for gas-phase detection of explosives with ppb sensitivity.
• Development of modules for water-phase detection of explosives with sub-ppm sensitivity.
• Development of a small form factor low power gamma radiation sensor with <10% energy resolution and an energy range of 60keV to 2MeV.
• Development of an intelligent learning network, using chemometric algorithm to teach itself to detect explosives and ignore interferents.
Technology
The purpose of the CommonSense project is to provide greater protection to EU citizens from the threats posed by the use of IED manufacturing facilities in urban environments from criminal, terrorist and other organisations, without the use of intrusive surveillance and tracking techniques. The partners will develop a series of different sensor types for detection of chemical explosives and radioactive materials, together with chemometric data processing algorithms to recognise trace amounts of explosives, and differentiate them from other interfering compounds. The schematic below shows an example of one possible means of deploying such a sensor network to better protect EU citizens from the threat of IEDs.
In order to achieve these ambitious objectives, work plan for the CommonSense project is divided into five complementary technical work packages:
Design and Specification
At the start of the project, the partners will specify target IED analytes, detection limits and test conditions relevant to end users. Specification of the common testing and benchmarking procedures, operating protocols, network architectures and communications protocols will also be carried out.
Materials Development and Characterisation
A variety of novel molecular, polymeric and nanostructured sensor materials will be developed and characterised with respect to their optoelectrical and photophysical properties, especially their response to sub-ppb (gas phase) and sub-ppm (liquid) levels of explosive compounds.
Sensor Development
Development of the sensor modules will be carried out at separate partner sites for initial testing and characterisation. A variety of different electrical, opto-electrical and opto-electrochemical devices for gas- and water-phase detection of IED analytes will be developed. A series radiation detection modules will also be developed.
Software Development and Networking
This WP focuses on the development of the common network platform for control and communication of the sensor modules. Driver software for control and read-out from different sensor types will be done at partner sites prior to integration with the network and the chemometric “learning” algorithms.
Integration, Testing and Industrial Validation
The final WP focuses on the integration of the sensor modules and quantitative testing and validation of the performance of the sensor modules. The final testing and assessment will be carried out in a “real-world” proving ground.
These technical work packages are supported by two non-technical work packages focusing on dissemination & exploitation of project results and project management.
Project Results:
Progress Overview
1. Nanoelectrochemical sensors and packaged modules for solution-based detection of explosives
Nanoelectrochemical sensors and packaged modules were developed by Tyndall-UCC for solution-based detection and discrimination of military-grade nitroaromatic and nitroester explosives and related compounds (DNT, TNT, PETN) with detection limits below 10 ppb for TNT.
These species include military grade nitroaromatics, e.g. TNT, and military-grade nitroesters, e.g. PETN. Gold nanowire based electrochemical nanosensors were developed at silicon chip substrates. These sensors were fully characterised and exhibited significant enhancements in sensitivity (up to 1000 times) compared to commercially available electrochemical sensors. On-chip counter and reference electrodes integrated at the sensor devices ensured that the potentials at which different molecules were reduced and oxidised remained stable and extremely reproducible. Using un-modified pristine electrodes, limits of detection as low as 10 ng/L (i.e. 10 parts per billion) were achieved for TNT samples in water without the need for supporting electrolyte.
These nanosensors represent a significant improvement (two orders of magnitude) over the state of the art in electrochemical based sensing. In fact, analyte pre-concentration, dedicated laboratories and high-end analytical instrumental techniques, i.e. solid phase micro-extraction-HPLC-MS are normally required to achieve these detection limits. Unexpectedly, the sensitivity of the nanowires was such it was possible to differentiate between very similar nitroaromatic molecules (potential interfering compound) based on the position of the square wave voltammetry (SWV) electro-reduction peak. In fact, the specificity was sufficient to clearly identify between structural isomers (same molecule but different structural configuration) of DNT, i.e. 2,4 DNT and 2,6 DNT.
Chemometric pattern recognition algorithms developed by UNIMAN were then applied to data where it was possible to readily identify TNT from other electroreductive species. Concerning the nitroester explosive compound, e.g. PETN a clear electroreduction peak for the nitroester groups also occurred reproducibly at a specific applied potential. A limit of detection of 1 ppm was achieved for military-grade PETN (due to presence of excipients such as binders and plasticisers).
Work was also undertaken on military grade nitroamine explosive compounds, RDX, and a broad reductive peak also identified.
A field trial was undertaken at Tyndall using these modules with remote control via a wireless tablet. These sensor modules were fully packaged and could be incorporated into future wireless sensor network modules; see D5.4 pages 16-21 .
2. Fluorescence-based optical “turn-off” sensors for gas-phase detection of explosives
Fluorescence-based optical “turn-off” sensors and modules incorporating novel organic semiconductor materials prepared as gas-permeable membranes were developed. Detection limits down to ppb-level were achieved for gas-phase detection of TNT, DNT, NB and PETN (Del 3.2 v2.0 page 9) integrated into sensor module.
A variety of novel sensor materials based on π-conjugated organic oligomeric and polymeric systems that change their colour/luminescence properties in the presence of explosive analytes have been developed and characterised by Technion. Key to the sensitivity of these “turn-off” materials is the presence of nucleophile groups which chemically bind to explosive analytes, leading to their luminescence quenching. These novel materials, prepared as gas-permeable membranes, can thus accumulate explosive analytes, resulting it their high sensitivity.
The initial test system was built around a large metal test cell and used an off-the-shelf photodetector. This was a suitable test system but its large size and cost made it unsuitable for the final wireless network test module. However, it was possible to demonstrate sensitivity to a number of explosive compounds including TNT, DNT, NB and PETN, with ppb levels achieved for DNT and sub-ppb detection levels estimated for PETN.
To achieve this aim, a miniaturised sensor module was designed, which also contains a second detector to monitor the output of the LED. The prototype circuit boards carry pulse drive electronics for the UV LED, the transimpedance amplifier for the photodiode, and a PIC device with on-board A/D to provide digital data on fluorescence levels to the host lap top. The design makes use of a miniature blower to blow a high flow of air around both sides of the filter, meaning that flow would continue even if the filter was blocked over time with dust particles. DNT was detected at levels of 100 ppb
The “turn-off” fluorescence sensors were integrated into the design of the sensor module that was developed for the final “real-world” trials by ALPHASENSE. The wireless boxes were able to power and control a range of different sensors and communicate to the central hub using the communications protocols established by Thales. A full electronic and mechanical specification for the integration of the different sensors onto a wireless platform was produced. This specification included interface electronics, communication sub-system and local power supply. It also defined the mechanical enclosure and air flow and thermal management. The modules accept the following sensors: two four channel OFET sensors, turn-off fluorescent detectors, Photo ionisation detector, electrochemical detectors for ammonia, NOx, hydrogen peroxide and sulphur dioxide; temperature and humidity sensors, GPS and anti-tamper devices.
3. Integrated multi-sensor modules for detection of explosives (with wireless control and monitoring).
Integrated multi-sensor modules were also developed for final testing incorporating photoionisation detectors (100 ppb for DNT, 78 ppb for TATP), electrochemical sensors (200 ppb for TATP), fluorescence turn-off sensors and organic field-effect transistor sensors. (see D5.4 pages 4-15)
Performance assessment undertaken using real-time detection of explosives (gas and water) and radiation threats were also carried out in a “real-world” proving scenarios demonstrates the applicability and suitability of the COMMONSENSE Platform developed within the project. This following work was undertaken:
(i) selected sensor modules fully integrated into a sensor network demonstrating remote gas sensing of nitroaromatic and interferent species, were conducted on the CommonSense wireless boxes at the Alphasense facility in Braintree, UK. Testing was conducted by Alphasense and Thales with assistance from Atmospheric Sensors Limited (a sub-contractor of Alphasense).
(ii) Fully integrated electrochemical nanosensors modules demonstrating remote water testing of nitroaromatics, and peroxide. Testing was conducted at Tyndall National Institute, Ireland. Sensor operation and wireless data transfer was controlled using Android software
(iii) Fully integrated radiation sensors demonstrating remote sensing of barium, cobalt and caesium. Testing was conducted at Tyndall National Institute, Ireland.
4 network boxes were available for the testing. Each box contained the following sensors::
• 6 Alphasense A4 electrochemical sensors. These are designed for atmospheric monitoring but have high sensitivity. Although designed for specific analytes are able to detect other gases and vapours.
Sensors used were
- Hydrogen peroxide (2 sensors) EC1 and EC2
- Ammonia EC3
- Nitrogen dioxide EC4
- Sulphur dioxide EC5
- Nitrogen monoxide EC6
• Photoionisation detector (PID) Sensitive to a range of volatile organics at low concentrations. Non-specific but good sensitivity
• Organic Field Effect Transistor (OFET) Designed by university of Manchester
- Each chip has 4 OFETS
- Two boxes had 4 OFETS all sensitive to DNT
- Two boxes had 4 OFETS sensitive to DNT and 4 sensitive to TNT
• Fluorescent “turn-off” sensor developed by Technion
- Sensitive to DNT and other nitro and nitrogen containing materials
Prior to the testing the electrochemical sensors and boards were calibrated to obtain both the sensitivity of the sensors and also the zero currents. This would enable a concentration of an analyte in parts per billion (ppb) to be measured, i.e. a reading of 10 ppb from the NH3 sensor would be equivalent to 10 ppb of NH3 but does not necessarily equate to 10 ppb of the analyte. The boxes were designed such that a small fan would draw air from the test area into a flow channel containing the sensors. One of the boxes was modified such that analyte could be delivered by the Alphasense VOC system (this would also be compatible with the test system at BAM). Details of both of these systems are given in other deliverable reports.
The system deployed comprises several motes acting as network relays and 4 gas-phase sensors boxes integrating each one a mote; All this wireless sensors network were connected to the central data collection point computer (laptop) running Control and Command (C2) application providing graphical CBRNE situation awareness to the end-user. The system was designed to provide the sensor-mote communication path.
The system was deployed within the building on the ground floor and first floor to demonstrate usability of the system and good range for wireless coverage; see Figure 3. Integrated deployment assistance (wireless transmission power, link performance, error rate) was provided by the C2 application. Initial testing has been dedicated to validate sensor-mote communication whilst also testing the response and behaviour consistency of some of the electrochemical sensors and PID (e.g. temperature, acetone). The testing results quickly demonstrated working communication paths.
Subsequent testing was undertaken overnight (around 13 hours) with predefined readings transmission frequency (from 10sec. to 1 min). Sensors were stimulated by an Alphasense automated system injecting analytes (e.g. DNT). All sensor readings were saved on central data collection point computer hard disk as well on sensors boxes SD Card for post-analysis and comparison.
The network and central data collection point computer applications functioned well; 2.4 Ghz communication band performed correctly even in an industrial setting with metal embedded in the building fabric and clear evidence of interferences, fading etc. To optimise transmission, testing was performed at both 0db transmission power and +5db transmission power to assess several deployment configurations. System stability was confirmed; even with quite high rate readings transmission network was not overloaded (no data packet loss); combination of reasonable communication range and mesh network topology demonstrates the deployment flexibility and enables deployment even in large building. The C2 application with integrated deployment assistance is extremely useful to accelerate and ease indoor deployment.
4. Door-mounted, small-area radiation sensor modules
Door-mounted, small-area radiation sensor modules were developed and enabled detection of 112 kBq sources (60Co, 137Cs) at a stand-off distance of 32 cm and a 30 kBq source (133Ba) at a distance of 16.5 cm. (D3.1 D5.1 D5.4)
While the “gold standard” in terms of gamma detector performance are High Purity Germanium (HPGe) detectors, their cooling requirements prevent their use in compact battery powered devices. Semiconductor detectors offer excellent identification performance in a small form factor but with significantly less sensitivity and higher cost than scintillator and Photomultiplier Tube (PMT) gamma detectors currently deployed in airports and other high security locations. SensL and Tyndall-UCC in the COMMONSENSE project, have collaborated to develop a complete radiation detection module based on Si Photomultiplier (SiPM) technology with wireless connectivity.
Three new silicon fabrication processes were developed for readout of CsI(Tl) based radiation sensors which provided improved signal to noise, increased detection efficiency and reduced noise levels. A range of radiation sensors in different formats were developed and tested: single detectors (1.25mm 3mm and 6mm width), pixelated detector arrays (13mm x 13mm width) in ceramic packages, as well as a large area pixelated detectors arrays (50mm x 50mm width) with 64 readout channels per package.
Custom designed digital electronics was developed, with improved results compared to older analogue based designs. Comparison of multiple pixel sizes and different silicon process types was carried out to determine suitability to create sensor for radiation detection. Based on these experimental findings, a gamma radiation probe comprising a 5cm x 5cm x 7cm probe consisting of CSI(TI) scintillator and detector array assembled in light tight enclosure was developed. This sensor probe and electronics was tested using Cs-137, Co-60 and Ba-133 sources and initial results indicate that the 5cm probe should be capable of meeting the target sensitivity at room temperature.
Continued improvements in silicon processing and electronics have allowed the size of radiation sensor to be miniaturised while meeting the project requirements for detecting radiation sources in less than 1 second. The 1.25cm detector is capable of meeting the specification for Co60 and Ir192 sources while maintaining zero false alarms while for the Cs137 the false alarm rate is approximately 2.5%. A world leading energy resolution of 6.8% for Cs137 isotope was achieved with this new detection module while the module was characterised and successfully trialled in the field, detecting all the sources within the timeframes and distances specified in the specification (see D5.4 pages 22-29).
5. Secure, resilient and portable plug-n-play wireless communication network
A secure, resilient and portable plug-n-play wireless communication network based on IEEE 802.15.4 2006 standard and IETF standards (IPv6 & CoAP), also compliant with US DoD JPEO-CBD compliant Common CBRNE Sensor Interface (CCSI) software gateway for CBRNE sensors and applications interoperability was demonstrated and tested successfully. (DD4.1 D4.3 and 4.6 D5.4)
Firstly, discussion and analysis of the operational requirements and end-users scenarios developed in order to assess which requirements could be taken into account in the project and what kind of technologies and system solutions should work. Based on the requirements and these user scenarios, the very small form factor of the sensors and long term operation and the requirement to, so web services communication supported by mini PC was not appropriate. Taking into account the COMMONSENSE sensor constraints, a sensors network platform was specified as well as a complete infrastructure at system level and technical level with scenarios, architecture design and capabilities description; while at the same time investigating and putting in place a state of the art, available wireless network technologies covering very low power wireless node hardware called motes, embedded operating systems, network protocols and supporting applications for network management, deployment support.
The state of the art recommendation is a selection of the best trade-off between requirements from the project and available technologies, namely IEEE 802.15.4 for wireless communication, ARAGO Systems wireless hardware platform WisMote product, IETF 6LowPAN / IPv6 and CoAP for sensors data communication and administration.
The Common CBRNE Sensors Interface (CCSI) standard protocol (XML based) from US DoD JPEO-CBD has been chosen to provide the interoperability between CBRNE applications such as Control and Command (C2) and Chemometrics Data Fusion software and wireless sensors network. Concretely CCSI supports exchange of CBRNE data and commands in a formal and standardized way. A complete description of the interface between the wireless network and the Data Fusion Chemometrics software has been specified in D4.4.
Control and Command (C2) graphical application dedicated to a Police surveillance operator completes the wireless network; it allows CBRNE situation awareness, mission level configuration of the sensors network and high-level visualisation of the network behaviour in real-time. A very clear and positive benefit thanks to the architecture and technical choices such as CCSI is that integration work is kept limited both in term of time and efforts.
6. Intelligent “learning” network with world-first molecular discrimination
An intelligent “learning” network was developed, using chemometric algorithms to teach the system to detect explosives and ignore interferents. (D4.2 D4.4 D4.5 D4.6 D5.5)
It is clear that the diverse range of specialised sensors developed in the COMMONSENSE project requires a series of fast and efficient learning algorithms for effective data analysis. Unsupervised learning methods deal with the problem of trying to find hidden structure in unlabelled data. Principal component analysis (PCA) is an unsupervised multivariate data analysis technique with no a priori knowledge of experimental structure and is used for data dimensionality reduction and pattern recognition.
PCA was applied successfully to demonstrate that it can successfully distinguish between 5 closely related nitroaromatic compounds tested, with a concentration of less than 50 ppb and samplified in liquid phase, The compounds were exposed to and detected by 4 different devices, This was achievable due to the extreme sensitivity of the nanowire devices, which permits molecular discrimination due to differences in molecular energy levels – even between structural isomers of DNT. Consequently, CommonSense has delivered a World’s first in electrochemical detection.
Unsupervised cluster analysis: After generating the PCA, the first component (PC1) results were plotted against the second (PC2) principal component scores to visualize the discrimination between compounds. As each compound forms a clearly distinguishable cluster, the outcome of the analysis indicates that all compounds are successfully discriminated by PCA. In addition, two different samples of the TNT compound were tested (military-grade material obtained from the Irish Defence Forces and commercial sources). The results show that the two different TNT samples also form a unique cluster, although commercial samples showed a very large within-group variance.
The fact that the TNT samples form a unique cluster suggests that the device consistently detected the signal produced by this nitro compound. As PCA is an unsupervised technique and does not know the identity (nitro compound class) of the samples, these results suggest that the devices successfully and consistently detected within compound similarities and between compounds dissimilarities which, translated into spectral signals, enable accurate discrimination of nitro compounds.
Supervised cluster analysis. To improve group separation and find out how much of the differences among the compounds the devices detected, discriminant function analysis (DFA) was applied to the data. DFA is a supervised classification algorithm that discriminates groups using a priori knowledge of class membership (uses the knowledge of nitro compound identity). The computation of DFA involves the calculation of the inverse of a variance-covariance matrix which cannot be computed if the matrix is singular. The singularity of this matrix is avoided if variables are not collinear.
Therefore, after PCA, DFA was applied to the PCs computed by PCA (as the PCs are collinear) and this hybrid algorithm is called PC-DFA. In the present work, the PCs were used as the input variables for DFA and the results were validated by 1000 bootstrap cross-validations. Analysis of the data obtained from the nanowire sensors showed a similar trend to that observed by the PCA analysis where the clusters formed by compounds with similar chemical structures are near to each other. As new and more data become available, partners will further probe the models described here for the accuracy of compound identification and also propose new multivariate data analyses methods to examine the data.
Customised facilities established by BAM and Alphasense for validation of COMMONSENSE sensors and modules at sub 50 ppb levels
Customised facilities established by BAM and Alphasense for validation of COMMONSENSE sensors and modules developed for gas-phase detection of precursors, explosives and interferents at levels below 50 ppb. Del D5.2_V2.0 D5.3_ V3.0 & Supplemental report on Sensor Module Testing.
A custom made test rig for the testing of sensor prototypes response to the precursors, explosives and interferents identified at the start of the project. The system is designed to deliver vapours in the ppb and ppm range, with the potential of ppt, at a gas flow of 500 sccm (standard cubic centimetres per minute). This was built around 3 Owlstone VOC generators. These are marketed specifically for generating known concentrations of VOCs including explosive materials.
The Owlstone units were modified by equipping with 8, 250 and 500 sccm digital mass flow controllers. In combination with make-up air, this enables a large range of concentrations to be obtained from a limited number of permeation tubes. The gas delivery system provides air at controlled humidities using bespoke software, automating the testing. The make-up air is supplied from the zero air generated at Alphasense, which is scrubbed to remove trace organic and gaseous impurities.
Calibrated sources for DNT and the majority of precursors and interferents were obtained; the 3 channel unit is capable of delivery ppt concentrations and is also backed up by an existing 6 channel system that offered extra capacity for calibration and also delivery of VOCs in the ppb range
The test set-ups developed at BAM are capable of producing atmospheres with very low gas concentrations of explosives to be used during the verification of newly developed sensors. Response behaviour has been explored in order to know how the gas mixing units respond to changes in sample or gas temperature, flow rates, and changes of analytes. Atmospheres with low concentrations of explosives can be generated with high stability when time is allowed for the equilibrium to establish. Proportional down-mixing is also possible. In parallel several gas-chromatographic detection lines have been put into operation, such that calibration can take place as soon as an analyte is introduced in the reference chamber for sensor validation.
Both systems are in place and can be used to measure a range of gas phase sensors although it may be necessary for custom housings to be designed for each sensor type to hold the sensors/ sensor modules, feed the cables to the outside world.
BAM can currently offer stable, static low-concentrated gas phases of DNT, TNT, and RDX of 3 ppb. The concentrations have been calibrated by collecting the gas flow in an adsorption tube and subsequently eluting the collected material and determining via LC-UV analysis. Other analytes, with higher vapour pressure, can be prepared with a notice of 1 week in advance. For preliminary testing of sensors and in order to verify their operation almost saturated atmospheres of all common explosives can also be used from the head-space of their containments. From prior testing a DNT gas phase with a range from ~50 1000 ppb is possible to generate. The TNT gas phase has a lower concentration, 6±2 ppb at 40 °C. The generation of a higher concentrated and variable gas phase of RDX failed because of its low vapour pressure and the lack of methods to detect it in the lower ppt-range. A concentration of 3 ppb in gas phase requires 70-80 °C of the analyte. The same is true for PETN and even more for HMX. The gas phases of all explosives are analysed and verified by LC-UV or GC-MS after pre-concentration with adsorption tubes (air sampling tubes). In addition, tests with acetone, as a precursor or processing chemical, and TATP in a permeation tube were also carried out with the Owlstone apparatus concentrations between 300 and 1000 ppb acetone, and between 30 and 200 ppb TATP in air can be generated at 30 °C sample temperature. However, higher concentrations would be possible, when the sample temperature will be increased.
Validation of end-user testing site
An end-user testing site was established by PSNI with the aid of the UK Home Office Centre for Applied Science and Technology (CAST) (see D7.4 Annex 3, attached)
The Police Service of Northern Ireland has identified a suitable location for the “real-world” testing of the COMMONSENSE platform and sensor network. The site comprises of one floor, contained within this floor there are eight separate rooms including one toilet and shower room and one laundry room. It is believed that this structure will simulate a modern urban apartment/flat environment. The site has been evaluated by the Centre for Applied Science and Technology (CAST), a member of the project advisory board, and by members of Tyndall-UCC.
Following an on-site evaluation the possibility of effluent discharge sensors inside the sewage system adjacent to the building was proposed for the testing. CAST carried out Gas Dispersal Testing at this site to assess the effect on the ventilation system and to monitor the dispersal of gases. The report and results on this testing showed excellent results, with that site being very suitable for testing. Similar testing carried by other partners showed clear wireless connectivity for sensor placement throughout the site, while background radiation levels were also very suitable for testing.
Potential Impact:
The CommonSense project directly addresses key objectives of the Security theme, namely improved security for the citizens and infrastructure, and enhanced competitiveness for industry; by
Developing a multivariate sensing platform for the detection of IED factories that will enable rapid take-up and implementation by end users, and
Enhancing competitiveness of European industry within the global markets by strengthening multiple areas of commercial manufacturing and development.
Innovative and cost effective permanent monitoring in urban areas: The overall goal of the CommonSense project is to create and prototype a network of sensors, through the simultaneous and parallel development of novel materials, sensors and a wireless communications network. This adaptive network will employ chemometric data processing algorithms to “learn” to recognise trace amounts of explosives, and differentiate them from interferents. To achieve this ambitious goal, innovation in CommonSense will be driven along three axes: materials, sensors, integration & network development.
Materials Innovation: A series of novel low cost organic, polymeric and nanocrystalline materials with tuned optical and electronic properties and surface affinities will be synthesised and optimised to facilitate optical or electrical signal transduction methods exhibiting high sensitivity (ppb level detection).
Sensor Innovation: CommonSense project includes a number of different sensor technologies, all integrated seamlessly within a single integrated network. The different sensors will provide optical, electrical or electrochemical signal readout in real time.
Integration & network innovation: The project seeks to develop a series of stand-alone, independently operable sensor module with integrated power supplies, sampling hardware, novel sensors, control electronics and networking communications hardware. Full wireless connectivity for interrogation, operation and read-out within a single integrated, flexible and secure network with the capacity to interrogate sensors to monitor and also reconfigure them.
An intelligent, learning approach for analysis and decision making of results from remote sensor units will be developed. This approach will dramatically reduce roll-out, installation and maintenance costs, making them economically viable for deployment across selected areas of a European city. Furthermore, this approach provides a novel and flexible approach to standardisation, to allow the creation of a European common market for future mass transport security solutions, and interoperability of different security systems managed by different operators and/or between different EU countries.
Cost effective permanent monitoring in urban areas: The CommonSense sensor technology will provide test results that are as accurate and reliable as those obtained from laboratories while also demonstrating cost-effectiveness. These estimated costs are based on fabrication of greater than 1000 units using current costs available within the consortiums’ facilities. These costs would be significantly reduced when fabricating much larger device numbers at commercial quantities due to favourable economies of scale.
Enhancing competitiveness of European industry: The competitiveness of European industry depends largely on gaining new knowledge and new ways of integrating and exploiting existing knowledge. European sensing and security industry sectors will benefit economically by strategic enhancement of their global market position through the commercial development of advanced materials, processes and products. In this context, the CommonSense project will enhance competitiveness of European industry within the global markets by enhancing multiple areas of commercial manufacturing and development, including analytical instrumentation, integrated sensors, learning networks and communication sectors.
List of Websites:
Web: www.fp7projectcommonsense.eu
Email: info@fp7projectcommonsense.eu
The project has achieved significant improvements over the state of the art, both in terms of detection limits and novel analysis algorithms. Furthermore, the integration of sensor units into modules (both standalone and wireless-interfaced units) and subsequent testing emphasises the exploitation potential of the COMMONSENSE platform and its potential benefit for EU citizens and industry.
Project Highlights
1. Nanoelectrochemical sensors and packaged modules were developed for solution-based detection and discrimination of military-grade nitroaromatic and nitroester explosives and related compounds (e.g. DNT, TNT, PETN) with limits of detection (LoD) below 10 ppb for TNT.
2. Fluorescence-based optical “turn-off” sensors and modules incorporating novel organic semiconductor materials prepared as gas-permeable membranes were developed. Detection limits down to ppb-level were achieved for gas-phase detection of TNT, DNT, NB, PETN and RDX for lab-based sensors, 100 ppb for integrated modules.
3. Integrated multi-sensor modules (with wireless monitoring and control) were also developed for final testing. Modules incorporated photoionisation detectors (LoD ~ 100 ppb for DNT, 78 ppb for TATP), electrochemical sensors (LoD ~200 ppb for TATP), fluorescence turn-off sensors and organic field-effect transistor sensors.
4. Door-mounted, small-area radiation sensor modules were developed and enabled wireless detection of 112 kBq sources (60Co, 137Cs) at a stand-off distance of 32 cm and a 30 kBq source (133Ba) at a distance of 16.5 cm. These modules demonstrated world-leading energy resolution and low false-alarm rate.
5. A secure, resilient and portable plug-n-play wireless communication network has been developed and validated. The network, based on IEEE 802.15.4 2006 standard and IETF standards (IPv6 & CoAP), is also compliant with US DoD JPEO-CBD Common CBRNE Sensor Interface (CCSI) and software gateway for CBRNE sensors. Applications interoperability was demonstrated and tested successfully.
6. An intelligent “learning” network was developed, using chemometric algorithms to teach the system to detect explosives and ignore interferents, yielding a “world-first” in discrimination of molecular isomers.
7. Customised/bespoke facilities were established for validation of COMMONSENSE sensors and modules developed for gas-phase detection of precursors, explosives and interferents at levels below 50 ppb.
8. An end-user testing site was established by PSNI with the aid of CAST (UK Home Office Centre for Applied Science and Technology).
A project website has been established at www.fp7projectcommonsense.eu to publicise the goals of the project, introduce the consortium and disseminate progress to the wider scientific community, industry and the public – details on peer-reviewed publications and presentations are available on the project web site.
Project Context and Objectives:
Objectives:
The detection of improvised explosive device (IED) manufacturing facilities is crucial for the security of citizens, as well as infrastructures and utilities. The Madrid (2004) and London (2005) bombings, as well as more recent events in Stockholm (2010) and Oslo (2011) have highlighted the need to tighten national security in the face of ongoing threats. The growing number of threats to the security of EC citizens may be traced by following the emergence of the nomenclature used to describe these threats. Where once the main threats were described as NBC (nuclear, biological and chemical), the current threats posed by terrorists, insurgents and criminals are labelled as CBRNE (Chemical, biological, radiological, nuclear and explosive).
Radiological weapons (the ‘R’ in CBRNE) lend themselves perfectly to spreading fear and long-term contamination of buildings or areas. The addition of ‘E’ for explosives is the latest extension of the acronym, and reflects the growing concern about the use of IEDs in urban environments. With the growing number of potential threats, it is no longer possible (or preferable) to use a single detector technology to ensure the security of citizens. Common sense dictates that if no single sensor technology exists to detect all types of explosive compounds, we must use the established technologies in combination to fully protect EC citizens. The need clearly exists for a single distributed network, with a common interface and communications protocol, to manage and communicate with a variety of different sensor technologies, and use the combined sensor data to produce clear and unequivocal results with low false positive/negative readings.
The purpose of the COMMONSENSE project is to provide greater protection to EU citizens from the threats posed by improvised explosive device (IED) “bomb factories” in urban environments without the use of intrusive surveillance and tracking techniques. COMMONSENSE has developed a very low power, portable, resilient and plug-n-play wireless sensor network for integrated detection of explosives, precursors and radio-active species in gas and water. The partners will produce a series of novel organic, polymeric and nanocrystalline materials with tuned optoelectronic properties and surface affinities to be used as the active sensor elements. These elements will be incorporated into devices based on optical, electrical, and other readout mechanisms, for detection of airborne and waterborne analytes.
The key point in the use of such a variety of sensor technologies is that no one substance can act as an interferent to all of the sensors, thus reducing false positives and negatives. Eliminating the remaining false readings will be achieved through use of the chemometric algorithms in order to teach itself to recognise the “fingerprint” sensor response to different explosives types and ignore interfering compounds. The COMMONSENSE network is intended to be deployed in European urban environments pervasive network, using low-cost sensor modules in wireless communication with a central server, which manages remote access, control, operation and readout from this network. Development and implementation of the COMMONSENSE will therefore enhance the daily security of European citizens via continuously monitoring of their home and work environments, and alerting security forces and other responders to the presence of IED manufacturing facilities.
The main objectives of the COMMONSENSE project are as follows:
• Development of modules for gas-phase detection of explosives with ppb sensitivity.
• Development of modules for water-phase detection of explosives with sub-ppm sensitivity.
• Development of a small form factor low power gamma radiation sensor with <10% energy resolution and an energy range of 60keV to 2MeV.
• Development of an intelligent learning network, using chemometric algorithm to teach itself to detect explosives and ignore interferents.
Technology
The purpose of the CommonSense project is to provide greater protection to EU citizens from the threats posed by the use of IED manufacturing facilities in urban environments from criminal, terrorist and other organisations, without the use of intrusive surveillance and tracking techniques. The partners will develop a series of different sensor types for detection of chemical explosives and radioactive materials, together with chemometric data processing algorithms to recognise trace amounts of explosives, and differentiate them from other interfering compounds. The schematic below shows an example of one possible means of deploying such a sensor network to better protect EU citizens from the threat of IEDs.
In order to achieve these ambitious objectives, work plan for the CommonSense project is divided into five complementary technical work packages:
Design and Specification
At the start of the project, the partners will specify target IED analytes, detection limits and test conditions relevant to end users. Specification of the common testing and benchmarking procedures, operating protocols, network architectures and communications protocols will also be carried out.
Materials Development and Characterisation
A variety of novel molecular, polymeric and nanostructured sensor materials will be developed and characterised with respect to their optoelectrical and photophysical properties, especially their response to sub-ppb (gas phase) and sub-ppm (liquid) levels of explosive compounds.
Sensor Development
Development of the sensor modules will be carried out at separate partner sites for initial testing and characterisation. A variety of different electrical, opto-electrical and opto-electrochemical devices for gas- and water-phase detection of IED analytes will be developed. A series radiation detection modules will also be developed.
Software Development and Networking
This WP focuses on the development of the common network platform for control and communication of the sensor modules. Driver software for control and read-out from different sensor types will be done at partner sites prior to integration with the network and the chemometric “learning” algorithms.
Integration, Testing and Industrial Validation
The final WP focuses on the integration of the sensor modules and quantitative testing and validation of the performance of the sensor modules. The final testing and assessment will be carried out in a “real-world” proving ground.
These technical work packages are supported by two non-technical work packages focusing on dissemination & exploitation of project results and project management.
Project Results:
Progress Overview
1. Nanoelectrochemical sensors and packaged modules for solution-based detection of explosives
Nanoelectrochemical sensors and packaged modules were developed by Tyndall-UCC for solution-based detection and discrimination of military-grade nitroaromatic and nitroester explosives and related compounds (DNT, TNT, PETN) with detection limits below 10 ppb for TNT.
These species include military grade nitroaromatics, e.g. TNT, and military-grade nitroesters, e.g. PETN. Gold nanowire based electrochemical nanosensors were developed at silicon chip substrates. These sensors were fully characterised and exhibited significant enhancements in sensitivity (up to 1000 times) compared to commercially available electrochemical sensors. On-chip counter and reference electrodes integrated at the sensor devices ensured that the potentials at which different molecules were reduced and oxidised remained stable and extremely reproducible. Using un-modified pristine electrodes, limits of detection as low as 10 ng/L (i.e. 10 parts per billion) were achieved for TNT samples in water without the need for supporting electrolyte.
These nanosensors represent a significant improvement (two orders of magnitude) over the state of the art in electrochemical based sensing. In fact, analyte pre-concentration, dedicated laboratories and high-end analytical instrumental techniques, i.e. solid phase micro-extraction-HPLC-MS are normally required to achieve these detection limits. Unexpectedly, the sensitivity of the nanowires was such it was possible to differentiate between very similar nitroaromatic molecules (potential interfering compound) based on the position of the square wave voltammetry (SWV) electro-reduction peak. In fact, the specificity was sufficient to clearly identify between structural isomers (same molecule but different structural configuration) of DNT, i.e. 2,4 DNT and 2,6 DNT.
Chemometric pattern recognition algorithms developed by UNIMAN were then applied to data where it was possible to readily identify TNT from other electroreductive species. Concerning the nitroester explosive compound, e.g. PETN a clear electroreduction peak for the nitroester groups also occurred reproducibly at a specific applied potential. A limit of detection of 1 ppm was achieved for military-grade PETN (due to presence of excipients such as binders and plasticisers).
Work was also undertaken on military grade nitroamine explosive compounds, RDX, and a broad reductive peak also identified.
A field trial was undertaken at Tyndall using these modules with remote control via a wireless tablet. These sensor modules were fully packaged and could be incorporated into future wireless sensor network modules; see D5.4 pages 16-21 .
2. Fluorescence-based optical “turn-off” sensors for gas-phase detection of explosives
Fluorescence-based optical “turn-off” sensors and modules incorporating novel organic semiconductor materials prepared as gas-permeable membranes were developed. Detection limits down to ppb-level were achieved for gas-phase detection of TNT, DNT, NB and PETN (Del 3.2 v2.0 page 9) integrated into sensor module.
A variety of novel sensor materials based on π-conjugated organic oligomeric and polymeric systems that change their colour/luminescence properties in the presence of explosive analytes have been developed and characterised by Technion. Key to the sensitivity of these “turn-off” materials is the presence of nucleophile groups which chemically bind to explosive analytes, leading to their luminescence quenching. These novel materials, prepared as gas-permeable membranes, can thus accumulate explosive analytes, resulting it their high sensitivity.
The initial test system was built around a large metal test cell and used an off-the-shelf photodetector. This was a suitable test system but its large size and cost made it unsuitable for the final wireless network test module. However, it was possible to demonstrate sensitivity to a number of explosive compounds including TNT, DNT, NB and PETN, with ppb levels achieved for DNT and sub-ppb detection levels estimated for PETN.
To achieve this aim, a miniaturised sensor module was designed, which also contains a second detector to monitor the output of the LED. The prototype circuit boards carry pulse drive electronics for the UV LED, the transimpedance amplifier for the photodiode, and a PIC device with on-board A/D to provide digital data on fluorescence levels to the host lap top. The design makes use of a miniature blower to blow a high flow of air around both sides of the filter, meaning that flow would continue even if the filter was blocked over time with dust particles. DNT was detected at levels of 100 ppb
The “turn-off” fluorescence sensors were integrated into the design of the sensor module that was developed for the final “real-world” trials by ALPHASENSE. The wireless boxes were able to power and control a range of different sensors and communicate to the central hub using the communications protocols established by Thales. A full electronic and mechanical specification for the integration of the different sensors onto a wireless platform was produced. This specification included interface electronics, communication sub-system and local power supply. It also defined the mechanical enclosure and air flow and thermal management. The modules accept the following sensors: two four channel OFET sensors, turn-off fluorescent detectors, Photo ionisation detector, electrochemical detectors for ammonia, NOx, hydrogen peroxide and sulphur dioxide; temperature and humidity sensors, GPS and anti-tamper devices.
3. Integrated multi-sensor modules for detection of explosives (with wireless control and monitoring).
Integrated multi-sensor modules were also developed for final testing incorporating photoionisation detectors (100 ppb for DNT, 78 ppb for TATP), electrochemical sensors (200 ppb for TATP), fluorescence turn-off sensors and organic field-effect transistor sensors. (see D5.4 pages 4-15)
Performance assessment undertaken using real-time detection of explosives (gas and water) and radiation threats were also carried out in a “real-world” proving scenarios demonstrates the applicability and suitability of the COMMONSENSE Platform developed within the project. This following work was undertaken:
(i) selected sensor modules fully integrated into a sensor network demonstrating remote gas sensing of nitroaromatic and interferent species, were conducted on the CommonSense wireless boxes at the Alphasense facility in Braintree, UK. Testing was conducted by Alphasense and Thales with assistance from Atmospheric Sensors Limited (a sub-contractor of Alphasense).
(ii) Fully integrated electrochemical nanosensors modules demonstrating remote water testing of nitroaromatics, and peroxide. Testing was conducted at Tyndall National Institute, Ireland. Sensor operation and wireless data transfer was controlled using Android software
(iii) Fully integrated radiation sensors demonstrating remote sensing of barium, cobalt and caesium. Testing was conducted at Tyndall National Institute, Ireland.
4 network boxes were available for the testing. Each box contained the following sensors::
• 6 Alphasense A4 electrochemical sensors. These are designed for atmospheric monitoring but have high sensitivity. Although designed for specific analytes are able to detect other gases and vapours.
Sensors used were
- Hydrogen peroxide (2 sensors) EC1 and EC2
- Ammonia EC3
- Nitrogen dioxide EC4
- Sulphur dioxide EC5
- Nitrogen monoxide EC6
• Photoionisation detector (PID) Sensitive to a range of volatile organics at low concentrations. Non-specific but good sensitivity
• Organic Field Effect Transistor (OFET) Designed by university of Manchester
- Each chip has 4 OFETS
- Two boxes had 4 OFETS all sensitive to DNT
- Two boxes had 4 OFETS sensitive to DNT and 4 sensitive to TNT
• Fluorescent “turn-off” sensor developed by Technion
- Sensitive to DNT and other nitro and nitrogen containing materials
Prior to the testing the electrochemical sensors and boards were calibrated to obtain both the sensitivity of the sensors and also the zero currents. This would enable a concentration of an analyte in parts per billion (ppb) to be measured, i.e. a reading of 10 ppb from the NH3 sensor would be equivalent to 10 ppb of NH3 but does not necessarily equate to 10 ppb of the analyte. The boxes were designed such that a small fan would draw air from the test area into a flow channel containing the sensors. One of the boxes was modified such that analyte could be delivered by the Alphasense VOC system (this would also be compatible with the test system at BAM). Details of both of these systems are given in other deliverable reports.
The system deployed comprises several motes acting as network relays and 4 gas-phase sensors boxes integrating each one a mote; All this wireless sensors network were connected to the central data collection point computer (laptop) running Control and Command (C2) application providing graphical CBRNE situation awareness to the end-user. The system was designed to provide the sensor-mote communication path.
The system was deployed within the building on the ground floor and first floor to demonstrate usability of the system and good range for wireless coverage; see Figure 3. Integrated deployment assistance (wireless transmission power, link performance, error rate) was provided by the C2 application. Initial testing has been dedicated to validate sensor-mote communication whilst also testing the response and behaviour consistency of some of the electrochemical sensors and PID (e.g. temperature, acetone). The testing results quickly demonstrated working communication paths.
Subsequent testing was undertaken overnight (around 13 hours) with predefined readings transmission frequency (from 10sec. to 1 min). Sensors were stimulated by an Alphasense automated system injecting analytes (e.g. DNT). All sensor readings were saved on central data collection point computer hard disk as well on sensors boxes SD Card for post-analysis and comparison.
The network and central data collection point computer applications functioned well; 2.4 Ghz communication band performed correctly even in an industrial setting with metal embedded in the building fabric and clear evidence of interferences, fading etc. To optimise transmission, testing was performed at both 0db transmission power and +5db transmission power to assess several deployment configurations. System stability was confirmed; even with quite high rate readings transmission network was not overloaded (no data packet loss); combination of reasonable communication range and mesh network topology demonstrates the deployment flexibility and enables deployment even in large building. The C2 application with integrated deployment assistance is extremely useful to accelerate and ease indoor deployment.
4. Door-mounted, small-area radiation sensor modules
Door-mounted, small-area radiation sensor modules were developed and enabled detection of 112 kBq sources (60Co, 137Cs) at a stand-off distance of 32 cm and a 30 kBq source (133Ba) at a distance of 16.5 cm. (D3.1 D5.1 D5.4)
While the “gold standard” in terms of gamma detector performance are High Purity Germanium (HPGe) detectors, their cooling requirements prevent their use in compact battery powered devices. Semiconductor detectors offer excellent identification performance in a small form factor but with significantly less sensitivity and higher cost than scintillator and Photomultiplier Tube (PMT) gamma detectors currently deployed in airports and other high security locations. SensL and Tyndall-UCC in the COMMONSENSE project, have collaborated to develop a complete radiation detection module based on Si Photomultiplier (SiPM) technology with wireless connectivity.
Three new silicon fabrication processes were developed for readout of CsI(Tl) based radiation sensors which provided improved signal to noise, increased detection efficiency and reduced noise levels. A range of radiation sensors in different formats were developed and tested: single detectors (1.25mm 3mm and 6mm width), pixelated detector arrays (13mm x 13mm width) in ceramic packages, as well as a large area pixelated detectors arrays (50mm x 50mm width) with 64 readout channels per package.
Custom designed digital electronics was developed, with improved results compared to older analogue based designs. Comparison of multiple pixel sizes and different silicon process types was carried out to determine suitability to create sensor for radiation detection. Based on these experimental findings, a gamma radiation probe comprising a 5cm x 5cm x 7cm probe consisting of CSI(TI) scintillator and detector array assembled in light tight enclosure was developed. This sensor probe and electronics was tested using Cs-137, Co-60 and Ba-133 sources and initial results indicate that the 5cm probe should be capable of meeting the target sensitivity at room temperature.
Continued improvements in silicon processing and electronics have allowed the size of radiation sensor to be miniaturised while meeting the project requirements for detecting radiation sources in less than 1 second. The 1.25cm detector is capable of meeting the specification for Co60 and Ir192 sources while maintaining zero false alarms while for the Cs137 the false alarm rate is approximately 2.5%. A world leading energy resolution of 6.8% for Cs137 isotope was achieved with this new detection module while the module was characterised and successfully trialled in the field, detecting all the sources within the timeframes and distances specified in the specification (see D5.4 pages 22-29).
5. Secure, resilient and portable plug-n-play wireless communication network
A secure, resilient and portable plug-n-play wireless communication network based on IEEE 802.15.4 2006 standard and IETF standards (IPv6 & CoAP), also compliant with US DoD JPEO-CBD compliant Common CBRNE Sensor Interface (CCSI) software gateway for CBRNE sensors and applications interoperability was demonstrated and tested successfully. (DD4.1 D4.3 and 4.6 D5.4)
Firstly, discussion and analysis of the operational requirements and end-users scenarios developed in order to assess which requirements could be taken into account in the project and what kind of technologies and system solutions should work. Based on the requirements and these user scenarios, the very small form factor of the sensors and long term operation and the requirement to, so web services communication supported by mini PC was not appropriate. Taking into account the COMMONSENSE sensor constraints, a sensors network platform was specified as well as a complete infrastructure at system level and technical level with scenarios, architecture design and capabilities description; while at the same time investigating and putting in place a state of the art, available wireless network technologies covering very low power wireless node hardware called motes, embedded operating systems, network protocols and supporting applications for network management, deployment support.
The state of the art recommendation is a selection of the best trade-off between requirements from the project and available technologies, namely IEEE 802.15.4 for wireless communication, ARAGO Systems wireless hardware platform WisMote product, IETF 6LowPAN / IPv6 and CoAP for sensors data communication and administration.
The Common CBRNE Sensors Interface (CCSI) standard protocol (XML based) from US DoD JPEO-CBD has been chosen to provide the interoperability between CBRNE applications such as Control and Command (C2) and Chemometrics Data Fusion software and wireless sensors network. Concretely CCSI supports exchange of CBRNE data and commands in a formal and standardized way. A complete description of the interface between the wireless network and the Data Fusion Chemometrics software has been specified in D4.4.
Control and Command (C2) graphical application dedicated to a Police surveillance operator completes the wireless network; it allows CBRNE situation awareness, mission level configuration of the sensors network and high-level visualisation of the network behaviour in real-time. A very clear and positive benefit thanks to the architecture and technical choices such as CCSI is that integration work is kept limited both in term of time and efforts.
6. Intelligent “learning” network with world-first molecular discrimination
An intelligent “learning” network was developed, using chemometric algorithms to teach the system to detect explosives and ignore interferents. (D4.2 D4.4 D4.5 D4.6 D5.5)
It is clear that the diverse range of specialised sensors developed in the COMMONSENSE project requires a series of fast and efficient learning algorithms for effective data analysis. Unsupervised learning methods deal with the problem of trying to find hidden structure in unlabelled data. Principal component analysis (PCA) is an unsupervised multivariate data analysis technique with no a priori knowledge of experimental structure and is used for data dimensionality reduction and pattern recognition.
PCA was applied successfully to demonstrate that it can successfully distinguish between 5 closely related nitroaromatic compounds tested, with a concentration of less than 50 ppb and samplified in liquid phase, The compounds were exposed to and detected by 4 different devices, This was achievable due to the extreme sensitivity of the nanowire devices, which permits molecular discrimination due to differences in molecular energy levels – even between structural isomers of DNT. Consequently, CommonSense has delivered a World’s first in electrochemical detection.
Unsupervised cluster analysis: After generating the PCA, the first component (PC1) results were plotted against the second (PC2) principal component scores to visualize the discrimination between compounds. As each compound forms a clearly distinguishable cluster, the outcome of the analysis indicates that all compounds are successfully discriminated by PCA. In addition, two different samples of the TNT compound were tested (military-grade material obtained from the Irish Defence Forces and commercial sources). The results show that the two different TNT samples also form a unique cluster, although commercial samples showed a very large within-group variance.
The fact that the TNT samples form a unique cluster suggests that the device consistently detected the signal produced by this nitro compound. As PCA is an unsupervised technique and does not know the identity (nitro compound class) of the samples, these results suggest that the devices successfully and consistently detected within compound similarities and between compounds dissimilarities which, translated into spectral signals, enable accurate discrimination of nitro compounds.
Supervised cluster analysis. To improve group separation and find out how much of the differences among the compounds the devices detected, discriminant function analysis (DFA) was applied to the data. DFA is a supervised classification algorithm that discriminates groups using a priori knowledge of class membership (uses the knowledge of nitro compound identity). The computation of DFA involves the calculation of the inverse of a variance-covariance matrix which cannot be computed if the matrix is singular. The singularity of this matrix is avoided if variables are not collinear.
Therefore, after PCA, DFA was applied to the PCs computed by PCA (as the PCs are collinear) and this hybrid algorithm is called PC-DFA. In the present work, the PCs were used as the input variables for DFA and the results were validated by 1000 bootstrap cross-validations. Analysis of the data obtained from the nanowire sensors showed a similar trend to that observed by the PCA analysis where the clusters formed by compounds with similar chemical structures are near to each other. As new and more data become available, partners will further probe the models described here for the accuracy of compound identification and also propose new multivariate data analyses methods to examine the data.
Customised facilities established by BAM and Alphasense for validation of COMMONSENSE sensors and modules at sub 50 ppb levels
Customised facilities established by BAM and Alphasense for validation of COMMONSENSE sensors and modules developed for gas-phase detection of precursors, explosives and interferents at levels below 50 ppb. Del D5.2_V2.0 D5.3_ V3.0 & Supplemental report on Sensor Module Testing.
A custom made test rig for the testing of sensor prototypes response to the precursors, explosives and interferents identified at the start of the project. The system is designed to deliver vapours in the ppb and ppm range, with the potential of ppt, at a gas flow of 500 sccm (standard cubic centimetres per minute). This was built around 3 Owlstone VOC generators. These are marketed specifically for generating known concentrations of VOCs including explosive materials.
The Owlstone units were modified by equipping with 8, 250 and 500 sccm digital mass flow controllers. In combination with make-up air, this enables a large range of concentrations to be obtained from a limited number of permeation tubes. The gas delivery system provides air at controlled humidities using bespoke software, automating the testing. The make-up air is supplied from the zero air generated at Alphasense, which is scrubbed to remove trace organic and gaseous impurities.
Calibrated sources for DNT and the majority of precursors and interferents were obtained; the 3 channel unit is capable of delivery ppt concentrations and is also backed up by an existing 6 channel system that offered extra capacity for calibration and also delivery of VOCs in the ppb range
The test set-ups developed at BAM are capable of producing atmospheres with very low gas concentrations of explosives to be used during the verification of newly developed sensors. Response behaviour has been explored in order to know how the gas mixing units respond to changes in sample or gas temperature, flow rates, and changes of analytes. Atmospheres with low concentrations of explosives can be generated with high stability when time is allowed for the equilibrium to establish. Proportional down-mixing is also possible. In parallel several gas-chromatographic detection lines have been put into operation, such that calibration can take place as soon as an analyte is introduced in the reference chamber for sensor validation.
Both systems are in place and can be used to measure a range of gas phase sensors although it may be necessary for custom housings to be designed for each sensor type to hold the sensors/ sensor modules, feed the cables to the outside world.
BAM can currently offer stable, static low-concentrated gas phases of DNT, TNT, and RDX of 3 ppb. The concentrations have been calibrated by collecting the gas flow in an adsorption tube and subsequently eluting the collected material and determining via LC-UV analysis. Other analytes, with higher vapour pressure, can be prepared with a notice of 1 week in advance. For preliminary testing of sensors and in order to verify their operation almost saturated atmospheres of all common explosives can also be used from the head-space of their containments. From prior testing a DNT gas phase with a range from ~50 1000 ppb is possible to generate. The TNT gas phase has a lower concentration, 6±2 ppb at 40 °C. The generation of a higher concentrated and variable gas phase of RDX failed because of its low vapour pressure and the lack of methods to detect it in the lower ppt-range. A concentration of 3 ppb in gas phase requires 70-80 °C of the analyte. The same is true for PETN and even more for HMX. The gas phases of all explosives are analysed and verified by LC-UV or GC-MS after pre-concentration with adsorption tubes (air sampling tubes). In addition, tests with acetone, as a precursor or processing chemical, and TATP in a permeation tube were also carried out with the Owlstone apparatus concentrations between 300 and 1000 ppb acetone, and between 30 and 200 ppb TATP in air can be generated at 30 °C sample temperature. However, higher concentrations would be possible, when the sample temperature will be increased.
Validation of end-user testing site
An end-user testing site was established by PSNI with the aid of the UK Home Office Centre for Applied Science and Technology (CAST) (see D7.4 Annex 3, attached)
The Police Service of Northern Ireland has identified a suitable location for the “real-world” testing of the COMMONSENSE platform and sensor network. The site comprises of one floor, contained within this floor there are eight separate rooms including one toilet and shower room and one laundry room. It is believed that this structure will simulate a modern urban apartment/flat environment. The site has been evaluated by the Centre for Applied Science and Technology (CAST), a member of the project advisory board, and by members of Tyndall-UCC.
Following an on-site evaluation the possibility of effluent discharge sensors inside the sewage system adjacent to the building was proposed for the testing. CAST carried out Gas Dispersal Testing at this site to assess the effect on the ventilation system and to monitor the dispersal of gases. The report and results on this testing showed excellent results, with that site being very suitable for testing. Similar testing carried by other partners showed clear wireless connectivity for sensor placement throughout the site, while background radiation levels were also very suitable for testing.
Potential Impact:
The CommonSense project directly addresses key objectives of the Security theme, namely improved security for the citizens and infrastructure, and enhanced competitiveness for industry; by
Developing a multivariate sensing platform for the detection of IED factories that will enable rapid take-up and implementation by end users, and
Enhancing competitiveness of European industry within the global markets by strengthening multiple areas of commercial manufacturing and development.
Innovative and cost effective permanent monitoring in urban areas: The overall goal of the CommonSense project is to create and prototype a network of sensors, through the simultaneous and parallel development of novel materials, sensors and a wireless communications network. This adaptive network will employ chemometric data processing algorithms to “learn” to recognise trace amounts of explosives, and differentiate them from interferents. To achieve this ambitious goal, innovation in CommonSense will be driven along three axes: materials, sensors, integration & network development.
Materials Innovation: A series of novel low cost organic, polymeric and nanocrystalline materials with tuned optical and electronic properties and surface affinities will be synthesised and optimised to facilitate optical or electrical signal transduction methods exhibiting high sensitivity (ppb level detection).
Sensor Innovation: CommonSense project includes a number of different sensor technologies, all integrated seamlessly within a single integrated network. The different sensors will provide optical, electrical or electrochemical signal readout in real time.
Integration & network innovation: The project seeks to develop a series of stand-alone, independently operable sensor module with integrated power supplies, sampling hardware, novel sensors, control electronics and networking communications hardware. Full wireless connectivity for interrogation, operation and read-out within a single integrated, flexible and secure network with the capacity to interrogate sensors to monitor and also reconfigure them.
An intelligent, learning approach for analysis and decision making of results from remote sensor units will be developed. This approach will dramatically reduce roll-out, installation and maintenance costs, making them economically viable for deployment across selected areas of a European city. Furthermore, this approach provides a novel and flexible approach to standardisation, to allow the creation of a European common market for future mass transport security solutions, and interoperability of different security systems managed by different operators and/or between different EU countries.
Cost effective permanent monitoring in urban areas: The CommonSense sensor technology will provide test results that are as accurate and reliable as those obtained from laboratories while also demonstrating cost-effectiveness. These estimated costs are based on fabrication of greater than 1000 units using current costs available within the consortiums’ facilities. These costs would be significantly reduced when fabricating much larger device numbers at commercial quantities due to favourable economies of scale.
Enhancing competitiveness of European industry: The competitiveness of European industry depends largely on gaining new knowledge and new ways of integrating and exploiting existing knowledge. European sensing and security industry sectors will benefit economically by strategic enhancement of their global market position through the commercial development of advanced materials, processes and products. In this context, the CommonSense project will enhance competitiveness of European industry within the global markets by enhancing multiple areas of commercial manufacturing and development, including analytical instrumentation, integrated sensors, learning networks and communication sectors.
List of Websites:
Web: www.fp7projectcommonsense.eu
Email: info@fp7projectcommonsense.eu