We study a crowdsourcing setting where we need to infer the latent truth about a task given observed labels together with context in the form of a classifier score. We present Theodon, a hierarchical non-parametric Bayesian model, developed and deployed at Meta, that captures both the prevalence of label categories and the accuracy of labelers as functions of the classifier score. Theodon uses Gaussian processes to model the non-uniformity of mistakes over the range of classifier scores. For our experiments, we used data generated from integrity applications at Meta as well as public datasets. We showed that Theodon (1) obtains 1-4% improvement in AUC-PR predictions on items' true labels compared to state-of-the-art baselines for public datasets, (2) is effective as a calibration method, and (3) provides detailed insights on labelers' performances.

Reliable, secure, private, and fast communication in vehicular networks is extremely challenging due to the highly mobile nature of these networks. Contact time between vehicles is very limited and topology is constantly changing. Trusted communication in vehicular networks is of crucial importance because without trust, all efforts for minimizing the delay or maximizing the reliability could be voided. In this paper, we propose a trust-based framework for a safe and reliable information dissemination in vehicular networks. The proposed framework consists of two modules such that the first one applies three security checks to make sure the message is trusted. It assigns a trust value to each road segment and one to each neighborhood, instead of each car. Thus, it scales up easily and is completely distributed. Once a message is evaluated and considered to be trustworthy, our method then in the second module looks for a safe path through which the message is forwarded. Our frameworks are application-centric; in particular, it is capable of preserving traffic requirements specified by each application. Experimental results demonstrate that this framework outperforms other well-known routing protocols since it routes the messages via trusted vehicles.

In this paper, we propose a cross-layer design to make ambulatory health monitoring via body area networks (BAN) more reliable and robust. The proposed design builds on our centralized body area network access scheme (CBAS), a receiver-initiated medium access control (MAC) scheme that improves the visibility of a BAN in a coexistent environment, where diverse networks with various physical and MAC protocols share the radio spectrum. The design enhances CBAS by incorporating a network layer scheme that improves the packet delivery ratio (PDR), while minimizing the need for multihop cooperative transmissions; thus, packet delay is less compromised to achieve higher PDRs. The MAC layer provides the network layer with local information about the quality of on-body links to enable the BAN to identify the most reliable links in a distributed manner. Extensive experimental results are presented, which give insights on how the proposed cross-layer design improves PDR and packet delay. Results show the effectiveness of the proposed design which takes advantage of dynamic scheduling and multihop relays as warranted by the link conditions.

This paper proposes a robust spectrum sensing strategy to identify white-spaces in the unlicensed 2.4GHz ISM band where radio transmission technologies coexist. Implicitly, collocated networks have different levels of spectrum access priority due to their design parameters (e.g., transmission power, modulation and coding), such that a WLAN tends to achieve a higher access priority compared to a coexisting WPAN. Instead of using a predetermined radio channel in the ISM band, a device can perform dynamic spectrum access using a frequency agile radio to take advantage of spectrum holes that are found between transmissions of collocated networks, thus improving its opportunities to access the ISM band for interference-free transmissions. This paper proposes the dual spectrum sensing strategy for capturing spectrum holes that are wide enough to accommodate the transmissions of a device. Our method makes no assumption about the technologies (e.g., modulation schemes or power spectral density) of detected radio transmissions, and properly handles power variations in the form of gradually changing power values as well as sharp rise/fall edges in the detection of occupied parts of the spectrum. Field test results show that ample white-spaces exist to enable frequency-agile IEEE 802.15.4-based devices to access the ISM band opportunistically.

Public m-health is a new medical service under intensive development, which provides unobtrusive monitoring of peoples health conditions from anywhere at any time to enable detection of deteriorating health conditions before severe discomfort or disability occur. A key challenge in practical implementation of public m-health is the use of shared licensefree spectrum by Body Area Networks (BANs) to report sampled vital signs continuously and in real-time. A cognitive medium access control method called Centralized Body Area Network Access Scheme (CBAS) is proposed in this paper to reduce access delay in a BAN in the presence of coexistent systems. By opportunistic extraction of idle spaces from a pool of orthogonal channels, CBAS dynamically adjusts a BANs channel access pattern according to the current interference environment, and improves the BANs visibility among coexistent networks. Performance of a BAN under CBAS is analyzed by modeling the system as a preemptive-resume priority queue. Numerical and simulation results show that the queuing-delay and throughput of a BAN employing CBAS outperforms those of a BAN that utilizes a single channel statically, as channel access opportunities suffer less fragmentations and interruptions.

A learning strategy for distributed channel selection in Cognitive Radio networks is proposed. This strategy helps quality of service (QoS) provisioning such that competing secondary users cooperatively converge to their rank-optimal channels while channel availability statistics are initially unknown. By this convergence, collision reaches zero since users eventually work on their own orthogonal channels. The proposed learning strategy, kth-MAB, is inspired from the Multi-Armed Bandit problem but it converges to the kthbest arm. The rank-optimal channel for each user is identified based on the user's QoS demands. We believe that this learning and allocation policy provides a better level of QoS for secondary users since evaluation results represent order optimality in terms of the average throughput.

Body Area Networks (BANs), targeting everywhere anytime health monitoring, operate in license-exempt frequency bands. They require a cross layer design that jointly considers the requirements of (1) access delay to the medium and (2) packet delivery ratio (PDR). In this paper, we experimentally evaluate the PDR of a BAN under static and dynamic scheduling schemes. Investigations show that static multi-hop routing enhances PDR with the help of packet relaying, but at the cost of using more transmission slots in the shared medium, which may compromise access delay. To reduce this cost, we propose a dynamic scheduling approach that adapts to BAN topology changes resulting from body movements. Our extensive experimental results indicate that for BANs colloccated with other wireless systems, our dynamic multi-hop scheduling is mandatory for satisfaction of the PDR constraint.

The Industrial, Scientific and Medical (ISM) band is shared among multiple wireless networks such as Bluetooth, wireless local area networks and 802.15.4 low rate wireless personal area networks (LR-WPAN). There are some centralized protocols that may work in this coexistent medium. Since management of these methods is done centrally with the help of configuration messages, their performance is strictly tied to the reliable transmission of these data. Clearly, this required level of reliability is in doubt when these methods are implemented in a heterogeneous environment. In this paper we explore how the performance of these standards is affected when they confront with interference. For this reason, WPAN beacon-enabled mode is selected as a representative of the category of synchronization-based methods, and its performance in a co-located WLAN\WPAN medium is evaluated. Then, a rapid-response and robust interference coexistence scheme is proposed to mitigate the effect of interfering systems on WPAN's performance. Simulation results show that the proposed method can be efficiently used to recover an affected WPAN quickly, regardless of the amount of WLAN interference.

A hybrid sensor network can be utilized to transfer multiple traffics with diverse delay and loss requirements. To satisfy the worst-case delay and loss requirements the network has to service each category of traffic according to its characteristics and quality of service (QoS) requirements. The proposed algorithm in this paper combines a specific time framing pattern, to prevent congestion, with a medium access scheduling method. This method assigns timeslots proportionally to the importance and QoS requirements of the traffic, to differentiate between the traffics. Simulation results show that packets of a specific source are delivered to the sink with a bounded delay and that delay bounds are proportional to the priority levels. These extensive simulation results approve the effectiveness of this method for satisfying QoS requirements in sensor networks.