WP3
Demonstration & Dissemination
This project will utilize two testbeds already available from some of the partners of the COMONSENS consortium: a testbed for MIMO communications with several transmit and receive antennas, and a testbed for a wireless ad-hoc sensor networking consisting of motes with different sensors and where all the configuration, programming and monitoring of the network can be performed remotely (including a web interface).
Apart from simulating algorithms, it is necessary to gain operational experience and to address the various technological and practical problems that are specific to the targeted application constraints of the various devices involved (such as low power, low cost, etc). The main goal is to establish a realistic setup where different codes and techniques can be compared, analyzing in practice crucial issues such as robustness, reliability, scalability and self-organization, and providing a valuable feedback into the design of theories and algorithms. The testbeds will serve both to keep the project grounded and to facilitate the timely impact of our theoretical and algorithmic propositions. Furthermore, during the project, the operational functionalities of these testbeds will be also drastically increased, which will make it easier to do research even directly on the application domain.
Moreover, the project also intends to build demonstrators for applications such as distributed detection, domotics, and environmental monitoring. Emphasis will be given to enhancing the ability to manage critical scenarios including: a) emergency situations and critical events, b) indoor positioning to help elderly or disable people in domestic contexts, and c) monitoring of environmental parameters to control the quality of citizens' lives.
This workpackage is led by UC and UDC, and involes the following partners: CEIT, UPC, UPM, UVEG, UC3M, UPF, UVIGO and US. The activities contained in this workpackage are described in more detail below.
A3.1 - Demonstration of Cooperative Networks
The objective of this activity is the construction of a cooperative wireless network hardware demonstrator for the practical evaluation and assessment of cooperative communications schemes in realistic scenarios. This will require the design and development of novel wireless nodes because, in addition to transmit and receive, terminals in cooperative networks should be able to assist in the reliable delivery of information from source terminals to destination terminals. Cooperation poses novel challenges in the design of practical terminals since it requires stronger coordination between the physical and the medium access control. Also, terminals should incorporate wireless operational feedback channels for the routing algorithms to perform properly and for the exchange of CSI.
The construction of a cooperative wireless network necessarily requires the cooperation of several research groups. Different partners of the COMONSENS consortium already have experience in the development of hardware demonstrators for MIMO point-to-point communications. This expertise will be exploited by COMONSENS to expand the existing hardware demonstrators in three directions:
Incorporation of cooperation capabilities into the existing terminals.
Replication of terminals to construct a cooperative wireless network with a large number of nodes that allow defining a wide variety of multi-terminal scenarios.
Provision of remote access to the hardware demonstrator through a web server so all research teams involved in COMONSENS can remotely perform their experiments.
The responsible partners of this activity (UDC and UC) have already participated in the MIMESIS coordinated project (TEC2004-06451-C05-01) and will participate during the following three years in the MULTIMIMO project (TEC2007-68020-C04). In the context of these projects, each partner has constructed its own MIMO hardware platform in a fruitful and collaborative fashion. Within COMONSENS, both platforms will be upgraded and brought together to construct a versatile cooperative network testbed.
A3.2 - Demonstration of Wireless Sensor Networks
The aim of this activity is to demonstrate the applicability of the new concepts and methodologies developed in the previous WPs and their impact in applications of interest.
Distributed detection for acoustic-source localization. In this scenario, a target is to be detected by means of many surrounding sensors. This target may be a person needing assistance, a radio beacon, a fire, an intruder, a pollutant contaminating an area, etc. This leads to a standard binary hypothesis testing problem. Target presence is related to a magnitude that depends on how the event is observed by each sensor at its specific position relative to the event occurrence. In our demonstrator, the target will be a voice emitter periodically requesting help. Around it, several nodes with unexpensive microphones will try to detect the presence of the target.
A hostile setup will be arranged, where neither knowledge on the network topology nor target position and propagation characteristics are available. Therefore, robust local detection at the nodes and robust fusion must be accomplished with uncertain (usually low) local signal-to-noise ratios and with no guaranteed links among the nodes. The objective of this test is to show how distributed detection may perform close to optimally while being much more efficient from a power consumption standpoint because only a handful of bits are transmitted, as opposed to a raw analog signal. Our deployment will be based on standard nodes, IEEE 802.15.4 compliant, with standard communication interfaces.
Our demonstrator will show how different configurations affect the fusion rule evaluating the system level detection performance in terms of the signal-to-noise ratio and of the model assumptions (a priori information) according to:
Precise knowledge of the sensors and the target positioning, which enable local signal-to-noise ratio and local probability of false alarm to be calculated if proper signal attenuation level is provided. Conversely, we will evaluate how much performance degrades when such positioning information is lacking.
Local detection rules and quantification of the received signal at every local node ranging from raw data transmission to a single bit wherever signal levels are above a predetermined threshold. The optimization of this threshold, depending on the available CSI, will also be performed.
Domestic monitoring system. A subset of the proposed distributed inference algorithms (including machine learning techniques as well as detection, tracking and prediction methods) will be applied to the surveillance and care of people with disabilities in a domestic context. We will show the feasibility of identifying, locating and tracking persons and also of learning a person's typical behavior patterns.
The sensor network will consist of a number of heterogeneous devices that measure acoustic energy, range, temperature, humidity, light, and video. Data collected at the sensor nodes will be transmitted, by means of a wireless network, to a small number of data fusion centers (DFCs) where specialized algorithms (location and tracking, classification, detection of events, etc) will be run. Note that indoor location cannot rely on global navigation satellite systems since their performance is severely hampered in multipath radio propagation environments. For this reason, the combination of sensor networks and nonlinear tracking algorithms for indoor location and navigation has become an active research topic that is attracting both academic and industrial attention. Furthermore, it constitutes a good benchmark for the techniques developed in WP2.
Although each DFC will provide information useful by itself, we will implement a higher level of information fusion by integrating the outputs of individual DFCs in order to adaptively learn the behavioral patterns of the persons being monitored. Knowledge of such patterns can help trigger real-time alarms, such as accidents or security events, and also long-term warnings motivated by the early detection of functional disorders.
Environmental monitoring. The monitoring of meteorological processes with high spatio-temporal resolution in environments with a strong anthropologic pressure, through the use of a dense wireless network of atmospheric sensors (temperature, humidity, etc...), where the meteorological conditioning determines very importantly the patterns of reception of pollutants and the occurrence of intense and sharp episodes of pollution. The experimental information of the atmospheric environment, having a sufficient coverage and density such as the one that can be provided by a wireless sensor network, will achieve two main practical objectives: a) much deeper understanding of the processes governing the pollutant emitter-receiver interaction and b) design of efficient strategies for management and prevention of atmospheric pollution. The ceramic environment of Castellón (Spain) provides an ideal scenario for the development of novel technologies, such as wireless sensor networks, because of the dense distribution of industrial manufacturers in an aerial watershed with a very particular dynamics, in addition to the availability of additional complementary historical experimental measurements of other types and an important cumulative knowledge about the region.
Our objective here will be to tailor the design and implementation of wireless sensor networks to a particular and important environmental monitoring application. We will build and deploy an outdoor low cost, reliable and reactive large-scale network integrating motes with different processing and communication capabilities. The scenario that will be considered is pollution monitoring in industrialized environments We will deploy a prototype wireless sensor network in the watershed of Mijares river, Castellón, which is an area strongly affected by pollution due to the presence of several ceramic manufacturers. By means of soil moisture, water pressure, rain and temperature sensors, the effect of pollution in water and soil conditions will be monitored. The steps to be followed are:
Design, implementation and calibration of a heterogeneous wireless sensor network in a controlled small-scale environment to test the various communication protocols and co-operative algorithms. In this stage, extensive use will be made of a testbed already developed by the Group of Information and Communication Systems at University of Valencia, with available web access at http://moteserver.uv.es.
Deployment of the wireless sensor network in the actual physical environment with various adverse environmental conditions with appropriate physical protection for the motes. This will include re-calibration and re-programming mechanisms that take into account the heterogeneity and hierarchy in terms of computation and communication capabilities, i.e., clusters of simple sensing motes will be reprogrammed by more powerful micro-server mote platforms, which will themselves receive reprogramming commands from other higher level nodes.
Exploiting documented information about the dynamics of the systems (pollution and atmosphere) in order to perform high-level inference. This will enable the early discovery of unusual patterns and the detection of extreme events.
Optimization of the wireless sensor network deployment, incorporating self-organized reactive mechanisms to ensure: a) reliable delivery of the majority (if not all) of the environmental data and metadata, b) high-quality (scientifically valuable) measurements, c) complete data analysis and seamless data access and visualization through a middleware web server, and d) reliable and continuous operation over long periods of time.
A3.3 - Dissemination and Knowledge Transfer
A number of specific actions have been planned in order to disseminate the results achieved by the COMONSENS consortium:
Publications in journals and conferences. Journals explicitly included in Thomson's ISI Web of Knowledge will be primarily targeted, with special interest in the first quarter of journals with higher impact factor. Particularly fitting are journals such as IEEE Transactions on Information Theory, IEEE Transactions on Signal Processing, IEEE Transactions on Communications, IEEE Transactions on Wireless Communications, IEEE/ACM Transactions on Networking and IEEE Signal Processing Magazine. Our target will be to ensure, on average, the publication at least one paper in a high-impact journal per year for every two researchers participating in the COMONSENS project, apart from publishing a substantial number of papers also in the best international conferences. This amounts to publishing at least 30 top journal papers per year in total. Conference presentations will be restricted to those events with a strict and rigorous peer review process. Thus far, Spain's research activity in the field of COMONSENS has mostly come from individual activities or small groups, without any institutional backing or much international repercussion. COMONSENS will be instrumental to strengthen the collaboration between the fragmented Spanish research mass, clearing the path to a significant international visibility. The synergy between the consortium groups is expected to amount to much more that the sum of the individual efforts.
COMONSENS workshop. We plan to celebrate an annual workshop where each of the groups will have the opportunity to present their main results and to sketch their future lines of investigation. Specialized seminars given by invited speakers, industry panels, on-line demonstrations of the developed testbeds, and many other activities of interest for the scope of the consortium, will also be planned. The members of the COMONSENS consortium have ample experience organizing events such as the National Assembly of the International Union of Radio Science (URSI), the IST Mobile Communications Summit 2001, ESA's Fifth International Workshop on Digital Signal Processing Techniques Applied to Space Communications, Learning'00 or the Sixth Baiona Workshop on Signal Processing in Communications.
Participation in technical committees. Members of the COMONSENS consortium have participated or are currently involved in international committees and standardization organisms such as eMobility platform, 3GPP (3rd Generation Partnership Project), IEEE 802 Working Group, Communication Theory Technical Committee, Technical Program Committee. To this string of activities we can add affiliation to the Swiss National Science Foundation Research Center, evaluation committee membership for European FP6 and FP7 projects, and technical program membership in numerous international conferences. The results of the proposed research are likely to further increase the presence of Spanish scientists in such committees, particularly since the topics of this proposal are well aligned with the ICT strategic agenda proposed in FP7 and with the agenda of the eMobility platform .
Co-organized Master and Ph.D. courses. COMONSENS will emphasize the need to build a solid educational structure around the project's research topics. Although some specialized courses and seminars are already established within the partners of the consortium (e.g. the coordinated Master program in information and communication technologies for mobile networks or the Advanced Sciences of Modern Telecommunications (ASMT) Master), we have a more ambitious goal: to fully interconnect the human and technical resources within the consortium, promoting the mobility of researchers and the internationalization of the co-organized courses and reinforcing the current status of the Spanish educational framework in the field. It should be stated that most of the principal researchers involved in the COMONSENS consortium have held academic positions in prestigious universities such as Columbia University, California State University, the University of Southern California and EPFL. Moreover, it is envisaged that the research pursued by the consortium will give rise to a number of Ph.D. dissertations supervised by researchers belonging to partners of the COMONSENS network, as well as to the creation of new graduate advanced courses. In addition, researchers and/or professors of international standing will be invited to deliver seminars or specialized courses, and to participate as members of the evaluation committees in each of the PhD theses developed within this consortium.
Technology and knowledge transfer to interested companies. Since innovation is a fundamental premise in COMONSENS, a major effort will be made to render the results, designs and developments available to interested companies. Since one of the main goals of COMONSENS is to strengthen the collaboration with European technology companies, the consortium will establish an Industrial Liaison Program (ILP) as a driver for knowledge and technology transfer within COMONSENS and as focal point for interacting with industry. This ILP will remain a very important tool to attract additional funding from industry and other funding agencies, fostering cooperation between COMONSENS partners and companies, and channeling results produced within COMONSENS towards industry through events such as industry courses and open workshops. Eight first-rate corporations have already shown their interest in the Consortium by providing the letters of interest endorsed in Appendix D.
Web page. This is a support web page to gather and publish technical information about research activities and events. This web service will serve as a link among the different centers and groups belonging to COMONSENS, as well as a means to make the developments made within COMONSENS accessible to all sectors of the society. Special funding will be devoted to contracting a part-time technician that designs and maintains the COMONSENS web service.
