Conference Papers

Opportunistic primary-secondary spectrum sharing with a rotating radar

Saruthirathanaworakun R., Peha J.M., Correia L.M.

2012 International Conference on Computing, Networking and Communications, ICNC'12

2012

Abstract:

This paper considers opportunistic primary-secondary spectrum sharing when the primary is a rotating radar. A secondary device is allowed to transmit when its resulting interference will not exceed the radar’s tolerable level, perhaps because the radar’s directional antenna is currently pointing elsewhere, in contrast to current approaches that prohibit secondary transmissions if radar signals are detected at any time. We consider the case where a secondary system provides point-to-multipoint communications utilizing OFDMA technology in non-contiguous cells, as might occur with a broadband hotspot service, or a cellular system that uses spectrum shared with radar to supplement its dedicated spectrum. We show that even at a fairly small distance to a radar, extensive secondary transmissions are possible, although with some interruptions as the radar rotates. For example, at 35% of the distance at which secondary transmissions will not affect radar, on average, the achievable secondary data rates in down-and upstream are around 100% and 46% of the rate that will be achieved in dedicated spectrum, respectively. By evaluating quality of service, we find that spectrum shared with radar could be used efficiently for applications such as non-interactive video on demand, peer-to-peer file sharing, large file transfers, automatic meter reading, and web browsing, but not for applications such as real-time transfers of small files and VoIP.

Opportunistic sharing between rotating radar and cellular

Saruthirathanaworakun R., Peha J.M., Correia L.M.

IEEE Journal on Selected Areas in Communications

2012

Abstract:

This paper considers opportunistic primary-secondary spectrum sharing when the primary is a rotating radar. A secondary device is allowed to transmit when its resulting interference will not exceed the radar’s tolerable level, in contrast to current approaches that prohibit secondary transmissions if radar signals are detected at any time. We consider the case where an OFDMA based secondary system operates in non-contiguous cells, as might occur with a broadband hotspot service, or a cellular system that uses spectrum shared with radar to supplement its dedicated spectrum. It is shown that even fairly close to a radar, extensive secondary transmissions are possible, although with some interruptions and fluctuations as the radar rotates. For example, at 27% of the distance at which secondary transmissions will not affect the radar, on average, the achievable secondary data rates in down- and upstreams are around 100% and 63% of the one that will be achieved in dedicated spectrum, respectively. Moreover, extensive secondary transmissions are still possible even at different values of key system parameters, including cell radius, transmit power, tolerable interference level, and radar rotating period. By evaluating quality of service, it is found that spectrum shared with radar could be used efficiently for applications such as non-interactive video on demand, peer-to-peer file sharing, file transfers, automatic meter reading, and web browsing, but not for applications such as real-time transfers of small files and VoIP.

Optimal Action Sequence Generation for Assistive Agents in Fixed Horizon Tasks

Baraka, K. ; Melo, F. S. ; Couto, M. ; Veloso, M.

Autonomous Agents and Multi-Agent Systems

2020

Abstract:

Agents providing assistance to humans are faced with the challenge of automatically adjusting the level of assistance to ensure optimal performance. In this work, we argue that identifying the right level of assistance consists in balancing positive assistance outcomes and some (domain-dependent) measure of cost associated with assistive actions. Towards this goal, we contribute a general mathematical framework for structured tasks where an agent playing the role of a ‘provider’—e.g., therapist, teacher—assists a human ‘receiver’—e.g., patient, student. We specifically consider tasks where the provider agent needs to plan a sequence of actions over a fixed time horizon, where actions are organized along a hierarchy with increasing success probabilities, and some associated costs. The goal of the provider is to achieve a success with the lowest expected cost possible. We present OAssistMe, an algorithm that generates cost-optimal action sequences given the action parameters, and investigate several extensions of it, motivated by different potential application domains. We provide an analysis of the algorithms, including proofs for a number of properties of optimal solutions that, we show, align with typical human provider strategies. Finally, we instantiate our theoretical framework in the context of robot-assisted therapy tasks for children with Autism Spectrum Disorder (ASD). In this context, we present methods for determining action parameters based on a survey of domain experts and real child-robot interaction data. Our contributions unlock increased levels of flexibility for agents introduced in a variety of assistive contexts.

Optimal cost actuator/sensor placement for large scale linear time-invariant systems: A structured systems approach

Pequito S., Kar S., Aguiar A.P.

2013 European Control Conference, ECC 2013

2013

Abstract:

We address the problem of minimal cost actuator/ sensor placement for large scale linear time invariant (LTI) systems that ensures structural controllability/observability. In particular, for the dedicated actuator placement problem (i.e., each actuator can control only one state variable or dynamic component), we propose a design methodology that provides the optimal placement with minimal cost (with respect to a given placement cost functional), under the requirement that the system be structurally controllable. In addition of obtaining the global solution of the optimization problem, the methodology is shown to be implemented by an algorithm with polynomial complexity (in the number of state variables), making it suitable for large scale systems. By duality, the solution readily extends to the structural design of the corresponding sensor placement under cost constraints.

Optimal design of distributed sensor networks for field reconstruction

Pequito S., Kruzick S., Kar S., Moura J.M.F., Aguiar A.P.

European Signal Processing Conference

2013

Abstract:

The paper introduces a method by which to design the topology of a distributed sensor network that is minimal with respect to a communication cost function. In the scenario considered, sensor nodes communicate with each other within a graph structure to update their data according to linear dynamics using neighbor node data. A subset of sensors can also report their state to a central location. One physical interpretation of this situation would be a set of spatially distributed wireless sensors which can communicate with other sensors within range to update data and can possibly connect to a network backbone. The costs would then be related to transmission energy. The objective is to recover the vector of initial sensor measurements from the backbone outputs over time, which requires that the dynamics of the overall networked system be observable. The topology of the network is then determined by the nonzero elements of the optimal observable dynamics. The following text contributes an efficient algorithm for designing the optimal observable dynamics and the network topology for a given set of sensors and cost function, providing proof of correctness and example implementation.

Optimal design of observable multi-agent networks: A structural system approach

Pequito S., Rego F., Kar S., Aguiar A.P., Pascoal A., Jones C.

2014 European Control Conference, ECC 2014

2014

Abstract:

This paper introduces a method to design observable directed multi-agent networks, that are: 1) either minimal with respect to a communications-related cost function, or 2) idem, under possible failure of direct communication between two agents. An observable multi-agent network is characterized by agents that update their states using a neighboring rule based on directed communication graph topology in order to share information about their states; furthermore, each agent can infer the initial information shared by all the agents. Sufficient conditions to ensure that 1) is satisfied are obtained by reducing the original problem to the travelling salesman problem (TSP). For the case described in 2), sufficient conditions for the existence of a minimal network are shown to be equivalent to the existence of two disjoint solutions to the TSP. The results obtained are illustrated with an example from the area of cooperative path following of multiple networked vehicles by resorting to an approximate solution to the TSP.

Optimal Perception Planning with Informed Heuristics Constructed from Visibility Maps

Pereira T., Moreira A., Veloso M.

Journal of Intelligent & Robotic Systems volume 93

2019

Abstract:

In this paper we consider the problem of motion planning for perception of a target position. A robot has to move to a position from where it can sense the target, while minimizing both motion and perception costs. The problem of finding paths for robots executing perception tasks can be solved optimally using informed search. In perception path planning, the solution when considering a straight line without obstacles is used as heuristic. In this work, we propose a heuristic that can improve the search efficiency. In order to reduce the node expansion using a more informed search, we use the robot Approximate Visibility Map (A-VM), which is used as a representation of the observability capability of a robot in a given environment. We show how the critical points used in A-VM provide information on the geometry of the environment, which can be used to improve the heuristic, increasing the search efficiency. The critical points allow a better estimation of the minimum motion and perception cost for targets in non-traversable regions that can only be sensed from further away. Finally, we show the contributed heuristic with improvements dominates the base PA* heuristic built on the euclidean distance, and then present the results of the performance increase in terms of node expansion and computation time.

Optimal strategies for side-channel leakage in FCFS packet schedulers

Shintre S., Gligor V., Barros J.

IEEE International Symposium on Information Theory - Proceedings

2015

Abstract:

We examine the side-channel information leakage in first-come-first-serve (FCFS) packet schedulers. In this setup, an attacker aims to learn the packet arrival pattern of a private user that shares a FCFS packet scheduler with him, using the queuing delay information of his own packets. Under an information-theoretic metric for information leakage, we identify the optimal non-adaptive strategy for a given average probe rate of the attacker and report upto 1000% increase in information leakage compared to the attack strategy analyzed in the literature with the same average probe rate. The search for optimal strategies is reduced to linear programming, implying that the discovery of such strategies is in the domain of a real-world attacker.

Ordinal losses for classification of cervical cancer risk

Albuquerque T., Cruz R., Cardoso J.

PeerJ Computer Science

2021

Abstract:

Cervical cancer is the fourth leading cause of cancer-related deaths in women, especially in low to middle-income countries. Despite the outburst of recent scientific advances, there is no totally effective treatment, especially when diagnosed in an advanced stage. Screening tests, such as cytology or colposcopy, have been responsible for a substantial decrease in cervical cancer deaths. Cervical cancer automatic screening via Pap smear is a highly valuable cell imaging-based detection tool, where cells must be classified as being within one of a multitude of ordinal classes, ranging from abnormal to normal. Current approaches to ordinal inference for neural networks are found to not sufficiently take advantage of the ordinal problem or to be too uncompromising. A non-parametric ordinal loss for neuronal networks is proposed that promotes the output probabilities to follow a unimodal distribution. This is done by imposing a set of different constraints over all pairs of consecutive labels which allows for a more flexible decision boundary relative to approaches from the literature. Our proposed loss is contrasted against other methods from the literature by using a plethora of deep architectures. A first conclusion is the benefit of using non-parametric ordinal losses against parametric losses in cervical cancer risk prediction. Additionally, the proposed loss is found to be the top-performer in several cases. The best performing model scores an accuracy of 75.6% for seven classes and 81.3% for four classes.