Fine-tuning large language models (LLMs) to align with user preferences is challenging due to the high cost of quality human annotations in Reinforcement Learning from Human Feedback (RLHF) and the generalizability limitations of AI Feedback. To address these challenges, we propose RLTHF, a human-AI hybrid framework that combines LLM-based initial alignment with selective human annotations to achieve full-human annotation alignment with minimal effort. RLTHF identifies hard-to-annotate samples mislabeled by LLMs using a reward model's reward distribution and iteratively enhances alignment by integrating strategic human corrections while leveraging LLM's correctly labeled samples. Evaluations on HH-RLHF and TL;DR datasets show that RLTHF reaches full-human annotation-level alignment with only 6-7% of the human annotation effort. Furthermore, models trained on RLTHF's curated datasets for downstream tasks outperform those trained on fully human-annotated datasets, underscoring the effectiveness of RLTHF.

Pico-satellite (picosat) constellations aim to become the de facto connectivity solution for Internet of Things (IoT) devices. These constellations rely on a large number of small picosats and offer global plug-and-play connectivity at low data rates, without the need for Earth-based gateways. As picosat constellations scale, they run into new bottlenecks due to their traditional medium access designs optimized for single (or few) satellite operations. We present CosMAC - a new constellation-scale medium access and scheduling system for picosat networks. CosMAC includes a new overlap-aware medium access approach for uplink from IoT to picosats and a new network layer that schedules downlink traffic from satellites. We empirically evaluate CosMAC using measurements from three picosats and large-scale trace-driven simulations for a 173 picosat network supporting 100k devices. Our results demonstrate that CosMAC can improve the overall network throughput by up to 6.5X over prior state-of-the-art satellite medium access schemes.

A recent development emanating from RFID technology is Near Field Communication (NFC). Basically, NFC is a popular short range (<10 cm) wireless communication technology with applications in areas sensitive to security and privacy concerns such as contactless payment. Since NFC communications require very close proximity between two communicating devices (e.g., a smartcard and a terminal), it is generally believed that Man-in-the-Middle (MITM) attacks are practically infeasible here. Contrasting this belief, in this paper, we successfully establish MITM attack in NFC communications between a passive tag and an active terminal. We carefully present physical fundamentals of the attack, our engineering design, and results of successful attack implementation. Subsequently, we present the practical applicability of our MITM attack that exploits a potential vulnerability in EMV based contactless payment protocol, which arises due to separation between card authentication and transaction authorization phases. We demonstrate how an attacker can compromise the integrity of a contactless payment using a malicious MITM card, and also present multiple attack/victim scenarios to analyze different types of impacts of our attack. Further, we conduct rigorous experimental studies to analyze both hardware and practical ramifications of our attack. Finally, we propose a mechanism to detect the MITM attack based on experimental analysis that demands no additional hardware.

In the context of developed countries, a considerable amount of research studies have been done to make Human-Computer Interaction (HCI) as natural and intuitive as possible for the public computing devices. Given the proliferation of public computing devices in the developing regions, a context-aware HCI design, focusing the low-literate and novice users from these regions, has the utmost importance. Taking the hygiene issue with the usage of touch-based public computing devices into account, in-air hand gestures are considered as one of the most natural forms of non-verbal interaction in this regard. In this paper, we present Duity -- a low-cost finger-count based hand gesture recognition system -- which reliably tackles the hurdle faced by novice and low-literacy populations. Exploiting the reflective property of visible light spectrum, we develop a mechanism that can reliably count fingers and thus identify different gestures. The system is fine-tuned to such an extent that varying lighting conditions and environmental setups can not perturb the accurate identification of the gestures. Our claims are supported by a user evaluation, which takes multiple scenarios into accounts, confirming the robustness, usability, and acceptability of Duity.

The proliferation of Internet of Things (IoT) devices has led to the deployment of various types of sensors in the homes, offices, buildings, lawns, cities, and even in agricultural farms. Since IoT applications rely on the fidelity of data reported by the sensors, it is important to detect a faulty sensor and isolate the cause of the fault. Existing fault detection techniques demand sensor domain knowledge along with the contextual information and historical data from similar near-by sensors. However, detecting a sensor fault by analyzing just the sensor data is non-trivial since a faulty sensor reading could mimic non-faulty sensor data. This paper presents a novel primitive, which we call the Fall-curve - a sensor's voltage response when the power is turned off - that can be used to characterize sensor faults. The Fall-curve constitutes a unique signature independent of the phenomenon being monitored which can be used to identify the sensor and determine whether the sensor is correctly operating.

The proliferation of Internet of Things (IoT) devices has led to the deployment of various types of sensors in the homes, offices, buildings, lawns, cities, and even in agricultural farms. Due to the diverse nature of IoT deployments and the likelihood of sensor failures in-the-wild, a key challenge in the design of IoT systems is ensuring the integrity, accuracy, and fidelity of sensor data.

Reading printed documents and writing on a paper pose a great challenge for visually-impaired people. Existing studies that attempt to solve these challenges are expensive and not feasible in low-income context. Moreover, these studies solve reading and writing problems separately. On the contrary, in this study, we propose FLight, a low-cost reading and writing system for economically less-privileged people. FLight uses ink-based Braille characters as the medium of textual representation. This helps in keeping a compact spatial representation of texts, yet achieving a low-cost status. Additionally, FLight utilizes a low-cost wearable device to enhance ease of reading by visually-impaired people. We conduct a participatory design and iterative evaluation involving five visually-impaired children in Bangladesh for more than 18 months. Our user evaluation reveals that FLight is easy-to-use, and exhibits a potential low-cost solution for economically less-privileged visually-impaired people.

The next generation ubiquitous wireless networks will comprise of mobile users moving between heterogeneous networks with terminals having multiple access interfaces and services. Vertical handover techniques deal with the scenario where a user seamlessly moves between different network access technologies. The most crucial issue in such environment is seamless connectivity. Earlier works demand for infrastructural modification with cumbersome implementations. Here, we propose an automated, transparent, seamless, hybrid decision model based easy-to-implement vertical handover mechanism between WiFi and cellular network. The implementation of the mechanism does not supplant the current infrastructure. We carry out experiments on a testbed to demonstrate the efficacy of our proposed scheme. Experimental results show that a handover-held communication slightly differs from a handoverless communication in terms of diversified network metrics.

A recent development emanating from the widely used RFID technology is Near Field Communication (NFC). Basically, NFC is a popular short range (<;10cm) wireless communication technology with applications in areas sensitive to security and privacy concerns including contact-less payment. Since NFC communications require very close proximity between two communicating devices (for example a smartcard and a reader), it is generally believed that Man-in-the-Middle (MITM) attacks are practically infeasible here. On the contrary to this general belief, in this paper, we successfully establish MITM attacks in NFC communications between a passive tag and an active reader. We present physical fundamentals of the attack, our engineering design, and results of our successful implementation. We also present practical impacts of the attack from the perspective of how a malicious user can leverage our MITM attack to compromise integrity of contact-less payment transactions. Finally, we present insights to combat the MITM attack in NFC communications towards the end of the paper.