|
Situation awareness superiority can provide tremendous strategic and tactical advantage to coalition forces over their adversary; it is especially challenging to achieve in cases of asymmetric warfare in urban contexts. Within the context of this technical area, we view networks of sensors (called sensor networks subsequently) and the data they generate as the powerful tools that aid in achieving situation awareness and supporting context-aware decision making and other high-level military operations.
Wireless sensor networks in military context are faced with unique deployment, operational and management challenges due to their operation in hostile, dynamic, unpredictable, resource-constrained, heterogeneous environments that reduce the chances of their long-term predictable operation. Given these operational realities, the overarching goal for the proposed research in technical area 3 (or, TA3) is to create the scientific underpinnings for the tools that will compose and deliver the best available, mission critical intelligence, surveillance and reconnaissance information extracted from a variety of sensory data in a timely, reliable and trustworthy manner to any level of decision making echelon that needs it, from the frontline combat elements in the theater of operation to the highest level of the command and control structure. To accommodate such an operating environment, solutions must consider information needs of a user and the ability of a network to delivery this information when both user needs and network operating conditions can change quickly and drastically with time.
To provide the aforementioned information in accordance with a security model and policies (as, for example, developed in TA2) to whomever needs it and whenever is needed will require addressing numerous issues, each giving rise to a set of research challenges. Specifically, providing the needed information will require the collaborative deployment and operation of sensor networks and data processing (including fusion) elements that process and deliver data from a collection of sensors of similar or different capabilities. The deployment of sensor networks should be content-guided. Warfighters should only need to specify what “high-level” information they need for their tasks and when, e.g., as by means developed in TA4, but not what data should be collected and/or from which sensors. Data and information must be qualified according to accuracy, reliability, security, urgency, and so on. Here, by deployment we imply not only the physical deployment and operation of sensor networks but also the logical deployment and operation of the necessary data gathering, processing, and delivery functionalities along an extended network of sensing, computing, and actuating platforms, including mobile platforms.
Maintaining the quality of the information gathered to support one or more higher-level functions, such as decision making, will require the development of the foundations of what “good” information means. This, in turn, will facilitate communicating to the data collecting elements (sensors) and services in the network (e.g., fusion points) in a quantifiable manner what kind of information will be sufficient for a task. In turn, this will enable sensors and data fusion elements to be deployed and configured in the best possible way to support a given set of tasks providing information of sufficiently high quality. Furthermore, repurposing and reconfiguration (including physical movement of sensors or reprocessing of data where applicable) of already deployed networks of sensors in the field could happen dynamically in response to current or in anticipation of future operational conditions (inc. unfolding events and environmental conditions) in order to continue providing, to the extent possible, the expected information at a sufficient quality as sensor and data fusion elements are compromised, incapacitated or repurposed.
To research and develop the tools that cope with the above challenges and, thus, reach our goal for TA3, we have organized the overall work into three projects focusing on: (i) the study of the quality of information as extracted from sensor nodes in the field and impacted by network transport and transformation services; (ii) the coordinated and collaborative deployment and configuration of network resources including sensors, platforms, forwarding nodes, and nodes that perform in-network processing; and (iii) the on-demand management of data and services that act on the data during mission planning and execution to enable availability of information as required by higher-level functions. In summary, the research projects in TA3 are:
Project 7 - Quality of Information in Sensor Network: This project aims to optimize and formally reason about the quality of information (QoI) obtained from sensor networks and presented to to decision makers. While the focus of the work is on QoI, we will extend this work to define and consider the Value of Information (VoI) as well. This work builds upon the QoI work done under the previous BPP, but significantly extends the work by separating QoI from VoI. Quality of Information (QoI) specifies how well the information may be used by a decision maker; VoI reflects the importance of the information in context. The goals of this work are to formalize QoI, to define and estimate the attributes of data that affect its QoI, and the model the information path from the source to the user to determine how transport and services in a network impact the QoI .
Project 8 - Task-Oriented Direction & Management of Sensor Networks: This project addresses the allocation of resources in mission-oriented sensor networks. Given mission descriptions, it is important to distill information needs, and then allocate resources so these needs may be met. Resources include sensors, and how they are shared among competing missions, pieces of information and data, and how they may be disseminated to users, and network resources, such as bandwidth. The goal of the work is to maximize the utility achieved by the sensor network over all of its missions over time accounting for competing missions of different priorities and realistic resource constraints. The utility is largely driven by metrics derived from QoI and to a certain extent VoI as defined in Project 7.
Project 9 - Agile sensor networks & data discovery: This project aims to develop theoretical models and algorithms for the management of distributed data and the services that operate on those data in a mobile ad hoc network (MANET) environment. For data access, a relational algebra is applied to manage data over ad hoc topologies that are highly dynamic; the work will leverage different connection strategies in the network to improve data retrieval performance. Likewise, service composition will formally model how services can be built, operated and dynamically updated in an environment of coalition partners operating with severely resource constrained and unreliable nodes.
Technical Area 3 may be viewed from the perspective of sensor information processing and information and resource management. The flow of data (and its derived information) is shown in Figure TA3-1. This Figure also shows, at a high level, the focus of each project. In this Figure we consider the collection of data from sensors which will be transformed into information. To satisfy information needs, a set of sensors {I} is selected. This selection is made based on the suitability of the sensor to capture the required information (e.g., the modality of the sensor, its location). It is likely that a set of sensors is required to fully meet the needs of a mission. It is also likely that multiple possible sets of sensors would satisfy the requirements of the mission. The selection of which specific sensors is therefore a function of the mission, the sensors, the state of the network, and all competing missions in the network. Once the sensors are selected, the best possible QoI and VoI (for an associated application) from the source are known.  Figure TA3-2. Information Processing Axis However, the data must be transferred over a network to its recipient. This transfer and any transformations that take place on the data, for example fusion, will increase or decrease the QoI. Network impairments (e.g., loss, latency) will cause degradation and network operations may cause an increase or decrease of QoI. To limit this loss services may be invoked in the network, such as fusion or compression. In this way the data being sent across the network is reduced in a controlled way, so that any information loss is predictable or in some cases avoided or reversed. The assignment of transport resources is made considering the type of data being transferred, how network characteristics will affect it, and considering the priority of missions. Likewise, the assignment of services is made with respect to its effect on the data and the availability of such services in the network. The manner in which data is received or disseminated is also impacted by the configuration of the network. Different network configurations will lend themselves to different models of data retrieval. Finally, as data crosses the network it is subject to security and policy imposed by coalition members. Thus, when data arrives at the destination, the QoI is now a function of the set of information sources {I}, the delivery capabilities of the network (D), the services in the network (Se) and the security (and policy) enforced in the network (Sc).
Project 7 is focused on developing technologies to define and quantify QoI and VoI for a representative set of military scenarios and applications . Thus, it mainly involves the impact of sensors on the ability to derive information. To this end Project 7 aims to characterize QoI and VoI in terms of data attributes, and model the impact of network delivery in the information content. Project 7 also examines the effect of compromised sensor integrity. Thus, in Figure TA3-1 1, Project 7 is shown primarily dealing with sensors, but in concert with Project 8, also being concerned with network behavior.
Project 8 is focused on resource allocation in terms of managing the network so that data may be delivered. Project 8 deals directly with arbitrating the selection of sensors to a task. Driving this selection is the QoI that may be achieved by the selected sensors. These QoI utility functions will be taken from Project 7 or developed collaboratively. Project 8 also addresses allocation of resources within the sensor network for transporting data, for example setting transmission rates, power, schedules and bandwidth. Finally, Project 8 investigates the optimal placement and impact of certain network operators, such as compression functions. Thus, in Figure TA3-1, it is shown straddling the sensor and the network, with some involvement in services.
Project 9 deals with data access and service composition. Data access must present a usable model to the user (e.g., making the network of sensors appear as a single uniform federated database) while providing good performance. Project 9 will leverage the configuration of the network when executing algorithms that map user needs and requests to methods of retrieving data. Project 9 also addresses the need to dynamically compose services to provide provably correct meta-services to an end user. These services can be fusion, compression, etc. The composition of services will be fed to Project 8 which is developing algorithms to allocate the correct resource to provide the service. Thus, Project 9 is shown mostly in the services area of Figure TA3-1, with an overlap with network configuration
Note that a similar figure could be drawn for information dissemination as opposed to collection. In this case, the information needs of a decision maker are portrayed as the need for a set of pieces of information. This information must then be delivered across a network, subject to all the constraints of the network and nodes operating within it, as well as the security rules and policies in the network. TA3 also addresses this case.
While the PIs in TA3 are assigned to individual projects, it is expected that they will work across project and TA boundaries frequently, will jointly publish papers, and define future research projects together. The research work in TA3 will be developed in a collaborative manner across the various technical areas through direct project involvement. Projects 7 and 8 have funded collaborators in TA1. Projects 8 and 9 have funded collaborators in TA4. TA3 also has various unfunded collaborators in TA1 (theory of wireless networks), TA2 (on policy) and TA4 (ontological reasoning). Additional information regarding project relationships, within TA3 and across technical areas, specifically TA1 and TA4, are discussed further in the subsequent project descriptions.
TA3 addresses hard problems A (MANETS), C (constrained resources), D (QoI) and G (information access). All three projects directly address MANETs (hard problem A). Project 7 models the impact of the network on QoI; Project 8 performs resource allocation and management of MANETs, and Project 9 leverages the MANET configuration to access data. Project 7 and Project 8 directly address operating with serious resource constraints (hard problem C). Project 7 explores the impact of degraded sensing abilities on QoI and Project 8 takes resource constraints into account when allocating sensors and transport resources. Project 7 and Project 8 also directly address QoI (hard problem D). Clearly, this is the focus of Project 7; Project 8 derives utility functions for sources based on QoI, and allocates resources to meet QoI targets. Project 9 also addresses information access (hard problem G). In fact, this is the focus of Project 9. For more details on the hard problems addressed by the projects in TA3, please see the detailed project descriptions that follow.
|
