Publisher Copyright: © 2026 IEEE. EC/HE/101071179/EU//SUSTAIN EC/HE/101099491/EU//HOLDEN

ACM TIOT launched a special issue on the theme of LLM Empowered Internet of Things, exploring the intersection of Large Language Models (LLMs) and the Internet of Things (IoT). As IoT continues to expand, advanced computational models are increasingly essential for processing and analyzing the massive data generated by interconnected devices. This special issue focuses on how LLMs can enhance IoT systems in several key areas. The second part of this special issue introduces the remaining six accepted papers that spans a board range of IoT scenarios from embedded and cyber-physical systems, human-centered applications, to IoT security.

ACM TIOT launched a special issue on the theme of LLM Empowered Internet of Things, exploring the intersection of Large Language Models (LLMs) and the Internet of Things (IoT). As IoT continues to expand, advanced computational models are increasingly essential for processing and analyzing the massive data generated by interconnected devices. This special issue focuses on how LLMs can enhance IoT systems in several key areas. These include improving context-aware perception and retrieval in complex IoT environments, applications of LLMs in human–computer interaction as well as applications of AI agents for IoT. The issue also highlights emerging trends in low-code and zero-code development for IoT programming, and the deployment of AI models at the edge. These research directions reflect the diverse and evolving landscape of IoT and LLM integration, offering innovative solutions to real-world challenges.

Gesture recognition for human machine interaction enhances the efficiency, safety, and usability of industrial and factory automation systems. We investigate hand-gesture recognition using battery-less body-worn reflective tags. Particularly, we propose two methods for hand gesture recognition using radio frequency identification (RFID). From backscattered signals we utilize in-phase and quadrature (IQ) constellation, as well as the phase. We convert extracted IQ samples into images and interprete them for gestures using a pre-trained VGG16. As a second approach we alternatively conduct pre-processing on the phase of the backscattered signals and propose Zero Crossing-Modified Derivative (ZCMD) for signal segmentation. Through signal resampling and wavelet denoising we mitigate undesired fluctuations introduced during this process, while retaining crucial signal characteristics. Subsequently, we integrate time-domain and frequency-domain features of the signals and train a random forest classifier based on these features to identify different gestures. Utilizing battery-free body-worn RFID tags, we are able to outperform a state-of-the art method and recognize four gestures with an accuracy of 81 % with the VGG16-based model. Employing phase, we achieve an accuracy of 94 %.

Recognition of human activities is crucial for enhancing safety, efficiency, and productivity within industrial and factory automation settings. This paper introduces a model for human activity recognition that leverages battery-less body-worn reflective antenna components. We perform preprocessing on both the backscattered phase and Received Signal Strength (RSS) signals. Independently and simultaneously, we extract features from phase and RSS signals using a feature extractor implementing a convolutional neural network (CNN). These features are then concatenated and fed into a fully connected (FC) layer employing the rectified linear unit (ReLU) activation function, followed by another FC layer utilizing a softmax function. This model, which merges extracted features from both phase and RSS, is termed late fusion model. We show that late fusion yields better performance than combining phase and RSS signals before feeding them into the neural network. By employing battery-free body-worn Radio frequency identification (RFID) tags, we surpass existing models, achieving an accuracy of 97.5% in recognizing five activities.

With the advancement of device capabilities, Internet of Things (IoT) devices can employ built-in hardware to perform machine learning (ML) tasks, extending their horizons in many promising directions. In traditional ML, data are sent to a server for training. However, this approach raises user privacy concerns. On the other hand, transferring user data to a cloud-centric environment results in increased latency. A decentralized ML technique, Federated learning (FL), has been proposed to enable devices to train locally on personal data and then send the data to a server for model aggregation. In these models, malicious devices, or devices with a minor contribution to a global model, increase communication rounds and resource usage. Likewise, heterogeneous data, such as non-independent and identically distributed (Non-IID), may decrease accuracy of the FL model. This paper proposes a mechanism to quantify device contributions based on weight divergence. We propose an outlier-removal approach which identifies irrelevant device updates. Client selection probabilities are computed using a Bayesian model. To obtain a global model, we employ a novel merging algorithm utilizing weight shifting values to ensure convergence towards more accurate predictions. A simulation using the MNIST dataset employing both non-iid and iid devices, distributed on 10 Jetson Nano devices, shows that our approach converges faster, significantly reduces communication cost, and improves accuracy.

Today’s deep neural networks (DNNs) are very accurate when trained on a large amount of data. However, suitable input might not be available or may require extensive data collection. Data sharing is one option to address these issues, but it is generally impractical because of privacy concerns or due to the problematic process of finding a sharing agreement. Instead, this work considers knowledge sharing by first exchanging the weights of pretrained DNNs and then applying transfer learning (TL). Specifically, it addresses the economics of knowledge sharing in AI services by taking a market-based approach. In detail, a model based on Fisher’s market is devised for optimal knowledge sharing, defined as the gain in inference accuracy from exchanging DNN weights. The proposed approach is shown to reach a market equilibrium and to satisfy important economic properties, including Pareto optimality. A technique for weight fusion is also introduced to merge acquired knowledge with the existing one. Finally, an extensive evaluation is conducted in a distributed intelligence scenario. The obtained results show that the proposed solution is efficient and that weight fusion with TL significantly increases inference accuracy compared to the original DNN, without the overhead of federated learning.

ACM TIOT launched its first special issue on the theme of wireless sensing for IoT. As an important component of the special issue and a novel practice of the journal, an online virtual workshop will be held, with presentations for each of the accepted articles. Welcome to join us for online discussion! Free registration is required for an attendee of the workshop. The zoom link will be shared to registered attendees before the workshop.

Ambient intelligence demands collaboration schemes for distributed constrained devices which are not only highly energy efficient in distributed sensing, processing and communication, but which also respect data privacy. Traditional algorithms for distributed processing suffer in Ambient intelligence domains either from limited data privacy, or from their excessive processing demands for constrained distributed devices. In this paper, we present Camouflage learning, a distributed machine learning scheme that obscures the trained model via probabilistic collaboration using physical-layer computation offloading and demonstrate the feasibility of the approach on backscatter communication prototypes and in comparison with Federated learning. We show that Camouflage learning is more energy efficient than traditional schemes and that it requires less communication overhead while reducing the computation load through physical-layer computation offloading. The scheme is synchronization-agnostic and thus appropriate for sharply constrained, synchronization-incapable devices. We demonstrate model training and inference on four distinct datasets and investigate the performance of the scheme with respect to communication range, impact of challenging communication environments, power consumption, and the backscatter hardware prototype.

We present Tesla-Rapture, a gesture recognition system for sparse point clouds generated by mmWave Radars. State of the art gesture recognition models are either too resource consuming or not sufficiently accurate for the integration into real-life scenarios using wearable or constrained equipment such as IoT devices (e.g., Raspberry PI), XR hardware (e.g., HoloLens), or smart-phones. To tackle this issue, we have developed Tesla, a Message Passing Neural Network (MPNN) graph convolution approach for mmWave radar point clouds. The model outperforms the state of the art on three datasets in terms of accuracy while reducing the computational complexity and, hence, the execution time. In particular, the approach, is able to predict a gesture almost 8 times faster than the most accurate competitor. Our performance evaluation in different scenarios (environments, angles, distances) shows that Tesla generalizes well and improves the accuracy up to 20% in challenging scenarios, such as a through-wall setting and sensing at extreme angles. Utilizing Tesla, we develop Tesla-Rapture, a real-time implementation using a mmWave Radar on a Raspberry PI 4 and evaluate its accuracy and time-complexity. We also publish the source code, the trained models, and the implementation of the model for embedded devices.

With the advance of machine learning and the Internet of Things (IoT), security and privacy have become critical concerns in mobile services and networks. Transferring data to a central unit violates the privacy of sensitive data. Federated learning mitigates this need to transfer local data by sharing model updates only. However, privacy leakage remains an issue. This paper proposes xMK-CKKS, an improved version of the MK-CKKS multi-key homomorphic encryption protocol, to design a novel privacy-preserving federated learning scheme. In this scheme, model updates are encrypted via an aggregated public key before sharing with a server for aggregation. For decryption, a collaboration among all participating devices is required. Our scheme prevents privacy leakage from publicly shared model updates in federated learning and is resistant to collusion between k < N − 1 participating devices and the server. The evaluation demonstrates that the scheme outperforms other innovations in communication and computational cost while preserving model accuracy.

Worker monitoring and protection in collaborative robot (cobots) industrial environments requires advanced sensing capabilities and flexible solutions to monitor the movements of the operator in close proximity of moving robots. Collaborative robotics is an active research area where Internet of Things (IoT) and novel sensing technologies are expected to play a critical role. Considering that no single technology can currently solve the problem of continuous worker monitoring, the article targets the development of an IoT multisensor data fusion (MDF) platform. It is based on an edge-cloud architecture that supports the combination and transformation of multiple sensing technologies to enable the passive and anonymous detection of workers. Multidimensional data acquisition from different IoT sources, signal preprocessing, feature extraction, data distribution, and fusion, along with machine learning (ML) and computing methods are described. The proposed IoT platform also comprises a practical solution for data fusion and analytics. It is able to perform opportunistic and real-time perception of workers by fusing and analyzing radio signals obtained from several interconnected IoT components, namely, a multiantenna WiFi installation (2.4-5 GHz), a sub-THz imaging camera (100 GHz), a network of radars (122 GHz) and infrared sensors (8-13 μm). The performance of the proposed IoT platform is validated through real use case scenarios inside a pilot industrial plant in which protective human-robot distance must be guaranteed considering latency and detection uncertainties.

Streaming 360° videos to a head-mounted display (HMD) client is challenging due to their high network resource consumption and computational load. This is due to the use of gaze point prediction or image saliency features from the field of view (FoV) since, in real-time scenarios, FoV extraction is computationally demanding. We propose a functional gaze prediction system that addresses these issues by relying on a tiling scheme for gaze prediction. We condition gaze point prediction on virtual reality (VR) content and long short-term memory (LSTM)-encoded eye movement history. Further, we encode image flow and saliency maps of RGB images via VGG16, using a convolutional neural network (CNN). Future gaze points are then predicted using a novel sinusoidal encoding technique. In experiments, our tile-based approach outperforms state-of-the-art FoV-based schemes in terms of computational load and predicted gaze position.

We describe a methodology to measure ballistocardiography (BCG) signals from the body surface, using body-worn digital accelerometers to extract medically relevant information for Pervasive Healthcare. We are able to measure measuring heart rate with an 95% accuracy as well as other cardiac metrics, such as the S1-S2 interval, deviating from ECG by only 1.3%. Our results show that BCG can be a viable alternative to an electrocardiogram to provide complementary information on the heart's condition in mobile and pervasive use cases. We further show that BCG information can be detected from arm as reliably as from chest, which is especially convenient for measuring from supine positions in Pervasive healthcare applications.

Artificial intelligence (AI) is among the most influential technologies to improve daily lives and to promote further economic activities. Recently, a distributed intelligence, referred to as a global brain, has been developed to optimize mobile services and their respective delivery networks. Inspired by interconnected neuron clusters in the human nervous system, it is an architecture interconnecting various AI entities. This paper models the global brain architecture and communication among its components based on multi-agent system technology and graph theory. We target two possible scenarios for communication and propose an optimized communication algorithm. Extensive experimental evaluations using the Java Agent Development Framework (JADE), reveal the performance of the global brain based on optimized communication in terms of network complexity, network load, and the number of exchanged messages. We adapt activity recognition as a real-world problem and show the efficiency of the proposed architecture and communication mechanism based on system accuracy and energy consumption as compared to centralized learning, using a real testbed comprised of NVIDIA Jetson Nanos. Finally, we discuss emerging technologies to foster future global brain machinelearning tasks, such as voice recognition, image processing, natural language processing, and big data processing.

Positive emotion is a pre-condition to any sales contract. Likewise, the ability to perceive the emotions of a customer impacts sales performance.To support emotional perception in buyer-seller interactions, we propose an audio-visual emotion recognition system that can recognize eight emotions: neutral, calm, sad, happy, angry, fearful, surprised, and disgusted. We reduced noise in audio samples and we applied transfer learning for image feature extraction based on a pre-trained deep neural network VGG16. For emotion recognition, we successfully obtained an audio emotion-recognition accuracy of 62.51% and 68% and video emotion-recognition accuracy of 97.13% and 97.77% on the Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) and Surrey Audio-Visual Expressed Emotion (SAVEE) datasets respectively. For the combination of the two models, our proposed merging mechanism without re-training achieved an accuracy of close to 100% on both datasets. Finally, we demonstrated our system for a customer satisfaction use case in a real customer-to-salesperson interaction using audio and video models, achieving an average accuracy of 78%.

We address the realization of the Findability, Accessibility, Interoperability, and Reusability (FAIR) data principles in an Internet of Things (IoT) application through a data transfer protocol. In particular, we propose an architecture for the Message Queuing Telemetry Transport (MQTT) protocol that validates, normalizes, and filters the incoming messages based on the FAIR principles to improve the interoperability and reusability of data. We show that our approach can significantly increase the degree of FAIRness of the system by evaluating the architecture using existing maturity indicators in the literature. The proposed architecture successfully passes 18 maturity indicators out of 22. We also evaluate the performance of the system in 4 different settings in terms the latency and dropped messages in a simulation environment. We demonstrate that the architecture not only improves the degree of FAIRness of the system but also reduces the dropped messages rate.

Recognition of the context of humans plays an important role in pervasive applications such as intrusion detection, human density estimation for heating, ventilation and air-conditioning in smart buildings, as well as safety guarantee for workers during human-robot interaction. Radio vision is able to provide these sensing capabilities with low privacy intrusion. A common challenge though, for current radio sensing solutions is to distinguish simultaneous movement from multiple subjects. We present an approach that exploits antenna installations, for instance, found in upcoming 5G technology, to detect and extract activities from spatially scattered human targets in an ad-hoc manner in arbitrary environments and without prior training of the multi-subject detection. We perform receiver-side beamforming and beam-sweeping over different azimuth angles to detect human presence in those regions separately. We characterize the resultant fluctuations in the spatial streams due to human influence using a case study and make the traces publicly available. We demonstrate the potential of this approach through two applications: 1) By feeding the similarities of the resulting spatial streams into a clustering algorithm, we count the humans in a given area without prior training. (up to 6 people in a 22.4 m2area with an accuracy that significantly exceeds the related work). 2) We demonstrate that simultaneously conducted activities and gestures can be extracted from the spatial streams through blind source separation.

Body movement is one of the most important indicators of sleep quality for elderly people living alone. Body movement is crucial for sleep staging and can be combined with other indicators, such as breathing and heart rate to monitor sleep quality. Nevertheless, traditional sleep monitoring methods are inconvenient and may invade users' privacy. To solve these problems, we propose a contactless body movement recognition (CBMR) method via WiFi signals. First, CBMR uses commercial off-the-shelf WiFi devices to collect channel state information (CSI) data of body movement and segment the CSI data by sliding window. Then, the context information of the segmented CSI data is learned by a bidirectional recurrent neural network (Bi-RNN). Bi-RNN can fuse the forward and backward propagation information at some point, and input it into a deeper independently recurrent neural network (IndRNN) with residual mechanism to extract the deeper features and capture the time dependencies of CSI data. Finally, the type of body movement can be recognized and classified by the softmax function. CBMR can effectively reduce data preprocessing and the delay caused by manually extracting features. The results of an experiment conducted on a complex body movement data set show that our method gives desirable performance and achieves an average accuracy of greater than 93.5%, which implies a prospect application of CBMR.

Gait has been proposed as a feature for mobile device pairing across arbitrary positions on the human body. Results indicate that the correlation in gait-based features across different body locations is sufficient to establish secure device pairing. However, the population size of the studies is limited and powerful attackers with e.g., capability of video recording are not considered. We present a concise discussion of security properties of gait-based pairing schemes including quantization, classification and analysis of attack surfaces, of statistical properties of generated sequences, an entropy analysis, as well as possible threats and security weaknesses. For one of the schemes considered, we present modifications to fix an identified security flaw. As a general limitation of gait-based authentication or pairing systems, we further demonstrate that an adversary with video support can create key sequences that are sufficiently close to on-body generated acceleration sequences to breach gait-based security mechanisms.

Smart glasses allow for gaze gesture passwords as a hands-free form of mobile authentication. However, pupil movements for password input are easily observed by attackers, who thereby can derive the password. In this paper we investigate closed-eye gaze gesture passwords with EOG sensors in smart glasses. We propose an approach to detect and recognize closed-eye gaze gestures, together with a 7 and 9 character gaze gesture alphabet. Our evaluation indicates good gaze gesture detection rates. However, recognition is challenging specifically for vertical eye movements with 71.2%-86.5% accuracy and better results for opened than closed eyes. We further find that closed-eye gaze gesture passwords are difficult to attack from observations with 0% success rate in our evaluation, while attacks on open eye passwords succeed with 61%. This indicates that closed-eye gaze gesture passwords protect the authentication secret significantly better than their open eye counterparts.

Contemporary personal mobile devices support a variety of authentication approaches, featuring different levels of security and usability. With cameras embedded in smart glasses, seamless, hands-free mobile authentication based on gaze is possible. Gaze authentication relies on knowledge as a secret, and gaze passwords are composed from a series of gaze points or gaze gestures. This paper investigates the concept of free-form mobile gaze passwords. Instead of relying on gaze gestures or points, free-form gaze gestures exploit the trajectory of the gaze over time. We collect and investigate a set of 29 different free-form gaze passwords from 19 subjects. In addition, the practical security of the approach is investigated in a study with 6 attackers observing eye movements during password input to subsequently perform spoofing. Our investigation indicates that most free-form gaze passwords can be expressed as a set of common geometrical shapes. Further, our free-form gaze authentication yields a true positive rate of 81% and a false positive rate with other gaze passwords of 12%, while targeted observation and spoofing is successful in 17.5% of all cases. Our usability study reveals that further work on the usability of gaze input is required as subjects reported that they felt uncomfortable creating and performing free-form passwords.

Popularity of mobile apps is traditionally measured by metrics such as the number of downloads, installations, or user ratings. A problem with these measures is that they reflect usage only indirectly. Indeed, retention rates, i.e., the number of days users continue to interact with an installed app, have been suggested to predict successful app lifecycles. We conduct the first independent and large-scale study of retention rates and usage trends on a dataset of app-usage data from a community of 339,842 users and more than 213,667 apps. Our analysis shows that, on average, applications lose 65% of their users in the first week, while very popular applications (top 100) lose only 35%. It also reveals, however, that many applications have more complex usage behaviour patterns due to seasonality, marketing, or other factors. To capture such effects, we develop a novel app-usage trend measure which provides instantaneous information about the popularity of an application. Analysis of our data using this trend filter shows that roughly 40% of all apps never gain more than a handful of users ( Marginal apps). Less than 0.1% of the remaining 60% are constantly popular ( Dominant apps), 1% have a quick drain of usage after an initial steep rise ( Expired apps), and 6% continuously rise in popularity ( Hot apps). From these, we can distinguish, for instance, trendsetters from copycat apps. We conclude by demonstrating that usage behaviour trend information can be used to develop better mobile app recommendations.

Recent advancements in vehicular technology have meant that integrated wireless devices such as Wi-Fi access points or bluetooth are deployed in vehicles at an increasingly dense scale. These vehicular network edge devices, while enabling in car wireless connectivity and infotainment services, can also be exploited as sensors to improve environmental and behavioural awareness that in turn can provide better and more personalised driver feedback and improve road safety. We present WiBot! a network-edge based behaviour recognition and gesture based personal assistant system for cars. WiBot leverages the vehicular network edge to detect distracted behaviour based on unusual head turns and arm movements during driving situations by monitoring radio frequency fluctuation patterns in real-time. Additionally, WiBot can recognise known gestures from natural arm movements while driving and use such gestures for passenger-car interaction. A key element of WiBot design is its impulsive windowing approach that allows start and end of gestures to be accurately identified in a continuous stream of data. We validate the system in a realistic driving environment by conducting a non-choreographed continuous recognition study with 40 participants at BMW Group Research, New Technologies and Innovation centre. By combining impulsive windowing with a unique selection of features from peaks and subcarrier analysis of RF CSI phase information, the system is able to achieve 94.5% accuracy for head-vs. arm movement separation. We can further confidently differentiate relevant gestures from random arm and head movements, head turns and idle movement with 90.5% accuracy.

The hypothesis of our research is that our body movements, posture and style of driving gives information about our behavioural state. When the driver is distracted by internal or external factors, his driving style changes. We detect these changes by utilising the non- intrusive, cheap and commercially off the shelf Wi-Fi sensors. We capture and analyse the fluctuations in Channel State Information (CSI) due to unusual body movements. The system detects distracted behaviour based on unusual head turns and arm movements during driving situations. This research is conducted in BMW Group Research, New Technologies and Innovation centre, Germany. We validate the hardware prototype by performing a human study of 40 participants, where the drivers are distracted by inducing unknown triggers. We capture the head and arm movements resulting in response of triggers. We introduce our denoising, phase correction and impulsive windowing technique to separate the human movements and distinguish between different activities. Combining this further with optimum time domain features from peaks and subcarrier analysis of CSI phase information, and applying multi- label classification techniques, the system is able to achieve 94.5% accuracy for head- vs. arm movement separation. We thus prove our hypothesis by spotting prolonged, unusual and frequent upper body movements.

We propose a platform for the integration of passive radio sensing and vision technologies into a cloud-IoT framework that performs real-time channel quality information (CQI) time series processing and analytics. Radio sensing and vision technologies allow to passively detect and track objects or persons by using radio waves as probe signals that encode a 2-D/3-D view of the environment they propagate through. View reconstruction from the received radio signals, or CQI, is based on real-time data processing tools, that combine multiple radio measurements from possibly heterogeneous IoT networks. The proposed platform is designed to efficiently store and analyze CQI time series of different types and provides formal semantics for CQI data manipulation-ontology models (OMs). Post-processed data can be then accessible to third parties via JSON-REST calls. Finally, the proposed system supports the reconfiguration of CQI data collection based on the respective application. The performance of the proposed tools are evaluated through two experimental case studies that focus on assisted living applications in a smartspace environment and on driver behavior recognition for in-car control services. Both studies adopt and compare different CQI manipulation models and radio devices as supported by current and future (5G) standards.

We present RFexpress! the first-ever network-edge based system to recognize emotion from movement, gesture and pose via Device-Free Activity Recognition (DFAR). With the proliferation of the IoT, also wireless access points are deployed at increasingly dense scale. in particular, this includes vehicular nodes (in-car WiFi or Bluetooth), office (Wlan APs, WiFi printer or projector) and private indoor domains (home WiFi mesh, Wireless media access), as well as public spaces (City/open WiFi, Cafes, shopping spaces). Processing RF-fluctuation at such edge-devices, enables environmental perception. In this paper, we focus on the distinction between neutral and agitated emotional states of humans from RF-fluctuation at the wireless network edge in realistic environments. In particular, the system is able to detect risky driving behaviour in a vehicular setting as well as spotting angry conversations in an indoor environment. We also study the effectiveness of edge-based DFAR emotion and activity recognition systems in real environments such as cafes, malls, outdoor and office spaces. We measure radio characteristics in these environments at different days and times and analyse the impact of variations in the Signal to Noise Ratio (SNR) on the accuracy of DFAR emotion and activity recognition. In a case study with 5 subjects, we then exploit the limits of edge-based DFAR by deriving critical SNR values under which activity and emotion recognition results are no longer reliable. In case studies with 8 and 5 subjects the system further could achieve recognition accuracies of 82.9% and 64% for vehicular and stationary wireless network edge in the wild (non-laboratory noisy environments and non-scripted, natural individual behaviour patterns).

Secure spontaneous authentication between devices worn at arbitrary locations on the same body is a challenging, yet unsolved problem. We propose BANDANA, the first-ever implicit secure device-to-device authentication scheme for devices worn on the same body. Our approach leverages instantaneous variations in acceleration patterns from the user's gait to extract always-fresh secure secrets. It enables secure spontaneous pairing of devices worn on the same body or interacted with. The method is robust against noise in sensor readings and active attackers.

While most RF-sensing approaches proposed in the literature rely on short-distance indoor point-to- point instrumentations, actual large-scale installation of RF sensing suggests the use of ubiquitously available cellular systems. In particular, the 5th generation of the wireless communication standard (5G) is envisioned as a universal communication means also for Internet of Things devices. In this paper, we investigate environmental perception capabilities in cellular systems, with a special focus on the upcoming 5G communications standard. In particular, we analyze the perception capabilities of existing cellular installations in the GSM band, which is expected to be used for 5G IoT. Our instrumentation exploits a passive system capitalizing on environmental RF-noise. In addition, utilizing a prototypical 5G system with 52 OFDM carriers over 12.48 MHz bandwidth at 3.45 GHz, we consider the impact of the number and choice of channels and compare the recognition performance with acceleration-based sensing. Our results in realistic settings with five subjects suggest that accurate recognition of activities and environmental situations can be a reliable implicit service of future 5G installations.

Exponential growth of media generated in online social networks demands effective recommendation to improve the efficiency of media access especially for mobile users. In particular, content, objective quality or general popularity are less decisive for the prediction of user-click behavior than friendship-conditioned patterns. Existing recommender systems however, rarely consider user behavior in real-life. By means of a large-scale data-driven analysis over real-life mobile Twitter traces from 15,144 users over a period of one year, we reveal the importance of social closeness related behavior features. This paper proposes CoCo, the first-ever behavior-inspired mobile social media recommender system to improve the media access experience. CoCo exploits representative behavior features via a latent bias based machine learning approach. Our comprehensive evaluation through trace-driven emulations on the Android app exposes a superior accuracy of 72.3%, with a small additional daily energy consumption of 1.3% and a monthly data overhead of 9.1MB.

Given the ubiquitous distribution of electronic devices equipped with a radio frequency (RF) interface, researchers have shown great interest in analyzing signal fluctuation on this interface for environmental perception. A popular example is the enabling of indoor localization with RF signals. As an alternative to active device-based positioning, device-free passive (DfP) indoor localization has the advantage that the sensed individuals do not require to carry RF sensors. We propose a probabilistic fingerprinting-based technique for DfP indoor localization. Our system adopts CSI readings derived from off-the-shelf WiFi 802.11n wireless cards which can provide fine-grained subchannel measurements in the context of MIMO-OFDM PHY layer parameters. This complex channel information enables accurate localization of non-equipped individuals. Our scheme further boosts the localization efficiency by using principal component analysis (PCA) to identify the most relevant feature vectors. The experimental results demonstrate that our system can achieve an accuracy of over 92% and an error distance smaller than 0.5m. We also investigate the effect of other parameters on the performance of our system, including packet transmission rate, the number of links as well as the number of principle components.

After a very successful edition of the IoT-App workshop, its second edition is conducted in conjunction with SenSys 2015. Again, the workshop succeeded in attracting a high number of high-quality submissions. The topics of the papers submitted feature most prominently urban sensing and smart home or city. Further directions are programming concepts for IoT devices as well as advances in machine learning, particularly deep learning for IoT devices.

We investigate the use of WiFi Received Signal Strength Information (RSSI) at a mobile phone for the recognition of situations, activities and gestures. In particular, we propose a device-free and passive activity recognition system that does not require any device carried by the user and uses ambient signals. We discuss challenges and lessons learned for the design of such a system on a mobile phone and propose appropriate features to extract activity characteristics from RSSI. We demonstrate the feasibility of recognising activities, gestures and environmental situations from RSSI obtained by a mobile phone. The case studies were conducted over a period of about two months in which about 12 hours of continuous RSSI data was sampled, in two countries and with 11 participants in total. Results demonstrate the potential to utilise RSSI for the extension of the environmental perception of a mobile device as well as for the interaction with touch-free gestures. The system achieves an accuracy of 0.51 while distinguishing as many as 11 gestures and can reach 0.72 on average for four more disparate ones.

We consider the detection of activities from non-cooperating individuals with features obtained on the radio frequency channel. Since environmental changes impact the transmission channel between devices, the detection of this alteration can be used to classify environmental situations. We identify relevant features to detect activities of non-actively transmitting subjects. In particular, we distinguish with high accuracy an empty environment or a walking, lying, crawling or standing person, in case-studies of an active, device-free activity recognition system with software defined radios. We distinguish between two cases in which the transmitter is either under the control of the system or ambient. For activity detection the application of one-stage and two-stage classifiers is considered. Apart from the discrimination of the above activities, we can show that a detected activity can also be localized simultaneously within an area of less than 1 meter radius.

We consider the recognition of activities from passive entities by analysing radio-frequency (RF)-channel fluctuation. In particular, we focus on the recognition of activities by active Software-defined-radio (SDR)-based Device-free Activity Recognition (DFAR) systems and investigate the localisation of activities performed, the generalisation of features for alternative environments and the distinction between walking speeds. Furthermore, we conduct case studies for Received Signal Strength (RSS)-based active and continuous signal-based passive systems to exploit the accuracy decrease in these related cases. All systems are compared to an accelerometer-based recognition system.

We propose to establish a secure communication channel among devices based on similar audio patterns. Features from ambient audio are used to generate a shared cryptographic key between devices without exchanging information about the ambient audio itself or the features utilized for the key generation process. We explore a common audio-fingerprinting approach and account for the noise in the derived fingerprints by employing error correcting codes. This fuzzy-cryptography scheme enables the adaptation of a specific value for the tolerated noise among fingerprints based on environmental conditions by altering the parameters of the error correction and the length of the audio samples utilized. In this paper, we experimentally verify the feasibility of the protocol in four different realistic settings and a laboratory experiment. The case studies include an office setting, a scenario where an attacker is capable of reproducing parts of the audio context, a setting near a traffic loaded road, and a crowded canteen environment. We apply statistical tests to show that the entropy of fingerprints based on ambient audio is high. The proposed scheme constitutes a totally unobtrusive but cryptographically strong security mechanism based on contextual information.

Artistic illustrations of important structures in fluid flow have a long-standing tradition and are appreciated as clearly perceivable, instructive, but still conveying all relevant information to the viewer. One important illustrative technique for such visualizations are cutaways. Currently cutaways are placed manually or using view-vector based approaches. We propose to optimize the visibility of important target features based on a degree-of-interest (DOI) function. The DOI is specified during interactive visual analysis, e.g., by brushing scatterplots. We show that the problem of placing cutaway boxes optimally is NP-hard in the number of boxes. To overcome this obstacle, we present an intelligent method to compute cutaways. Geometric cutaway objects are positioned using a view-dependent objective function which optimizes the visibility of all features. In order to approximate the optimal solution, we use a Monte Carlo method and exploit temporal coherence in dynamic scenes. Performance-critical parts are implemented on the GPU. The proposed method can be integrated easily into existing rendering frameworks and is general enough to be able to optimize other parameters besides cutaways as well. We evaluate the performance of the algorithm and provide a case study of vorticity visualization in a turbulent flow.

We introduce a passive system to recognise environmental situations. Differing from other RF-based approaches, our system has the advantage of neither installing a transmitter generating the signal nor equipping the monitored entities with any active component. When activities are performed, it consecutively samples ambient RF-signals, derived from a non-cooperating FM-radio source. Since changes in an environment impact the propagation of radio waves, this data implicitly contains information to distinguish environmental situations. We experimentally demonstrate the distinction of the situations 'empty room', 'opened door' and 'walking person' with an average accuracy of over 90%.

We introduce a novel activity recognition method based on the RF-signal originated from ambient FM radio source. For the purpose of classifying activities, we utilise a two stage approach which can initially distinguish between coarse-grained activities, then make further fine-grained recognition. Additionally, a study on features is conducted to investigate the most suitable combination to achieve the highest accuracy on the detection of activities. By comparing to a one stage classification process, the experimental results demonstrate the advantage of our designed approach.

We exploit activity recognition from RF-channels. Exceeding current studies, we discuss an implicit recognition scheme to compute context classifications with a network of wireless nodes. In particular, we propose a networked adhoc classification scheme that utilises the RF-features on the wireless channel among nodes as implicit inputs. Furthermore, we discuss the possibility to execute mathematical operations during transmission on the wireless channel. We present a data encoding which can be utilised to implicitly add, multiply, subtract or divide values during simultaneous transmission. In a simulation, we demonstrate the computation with a set of values by these implicit operations during transmission.

Context prediction is the task of inferring information about the progression of an observed context time series based on its previous behaviour. Prediction methods can be applied at several abstraction levels in the context processing chain. In a theoretical analysis as well as by means of experiments we show that the nature of the input data, the quality of the output, and finally the flow of processing operations used to make a prediction, are correlated. A comprehensive discussion of basic concepts in context prediction domains and a study on the effects of the context abstraction level on the context prediction accuracy in context prediction scenarios is provided. We develop a set of formulae that link scenario-dependent parameters to a probability for the context prediction accuracy. It is demonstrated that the results achieved in our theoretical analysis can also be confirmed in simulations as well as in experimental studies.

Intelligent Environments are currently implemented with standard WSN technologies using conventional connection-based communications. However, connection-based communications may impede progress towards IE scenarios involving high mobility or massive amounts of sensor nodes. We present a novel approach based on collective transmission for item level tagging using printed organic electronics, which implements robust, collective, approximate read-out of large numbers of simple tags. Our approach uses mechanisms for calculation by simultaneous transmission. We detail the collective transmission approach, discuss its implementation in the organic printed label scenario, and show first results of experiments conducted with our smart label test bed. We conclude with an outlook on the potential of collective transmission, and argue that collective transmission is a fundamental building block for realizing distributed intelligence.

In wireless sensor networks (WSN), collaboration is a way to improve the quality of data communication between sensor nodes with restricted resources in terms of memory, processing and energy storage. For receive collaboration, various array processing schemes such as receive beamforming and collaborative channel equalization (CCE) can be used for aggregating data received by each node in the network. The key challenge is the limitation on the number of nodes which can collaborate because of the increased computational load and memory demand when the multiple signals are aggregated. This problem arises when sensor nodes in a CDMA based WSN collaborate, although the low power property of CDMA technique makes it suitable for WSN applications. Here receive collaboration is investigated in CDMA networks using CCE as the collaboration algorithm. We present two novel distributed signal aggregation algorithms: partial and hierarchical aggregation, which distribute computational load and memory demands on collaborative nodes. The positive impacts of receive collaboration on the signal quality and reliability are confirmed experimentally in a WSN scenario using software radios. Then the requirements of collaborative reception using CCE combined with the novel aggregation methods in terms of computational and memory load, as well as energy consumption are evaluated. The results indicate that the distributed signal aggregation algorithms, especially hierarchical aggregation, have computational and memory requirements less than that of centralized CCE, providing greater flexibility and scalability which enables collaboration in WSNs on a larger scale than previously possible.

We study the feasibility of utilising the RF transceiver of a mobile device to establish some kind of situation awareness. Our results show that the analysis of channel characteristics can provide additional and sufficiently accurate context information. This situation awareness is cheap in the sense that it can be obtained from already ongoing communication in a network of nodes. In this paper we present results from a case study with several USRP software radios. We show how the presence, position and activity of persons in a room can be obtained exclusively from channel measurements.

We derive an asymptotically sharp bound on the synchronization speed of a randomized black box optimization technique for closed-loop feedback-based distributed adaptive beamforming in wireless sensor networks. We also show that the feedback function that guides this synchronization process is strong multimodal. Given this knowledge that no local optimum exists, we consider an approach to locally compute the phase offset of each individual carrier signal. With this design objective, an asymptotically optimal algorithm is derived. Additionally, we discuss the concept to reduce the optimization time and energy consumption by hierarchically clustering the network into subsets of nodes that achieve beamforming successively over all clusters. For the approaches discussed, we demonstrate their practical feasibility in simulations and experiments.

In this paper, a new type of collaboration in wireless sensor networks (WSN) is suggested that exploits array processing algorithms to improve the reception of a signal. For receive collaboration, the transmission power during intra-cluster transmissions decreases at the expense of increasing the inter-cluster communications. It is shown that, as a result of using receive collaboration, the destination node's power consumption and the network interference level decrease which considerably improve the data transmission performance and network life time. This method is applicable both for cluster based and non-cluster based WSNs. In order to show the feasibility of receive collaboration and also to evaluate its performance, an LS-CMA based channel equalization scheme is also simulated which is performed during cooperation between cluster nodes. The comparison of the output BER between random distributed and uniform linear distributed cases shows a good performance of receive collaboration.

Context prediction mechanisms proactively provide information on future contexts. Due to this knowledge novel applications become possible that provide services with proactive knowledge to users. The most serious problem of context prediction mechanisms lies in a basic property of prediction itself. A prediction is always a guess. Since erroneous predictions may cause the application to behave insufficiently, prediction errors have to be minimised. The accuracy of prediction is seriously affected by the reliability of the context data that is utilised by the method. We study two paradigms for context prediction and compare their potential prediction accuracy. We show that the designer of context prediction architectures has to choose wisely as to which prediction paradigm to follow in order to maximise the accuracy of the whole architecture. We also introduce a simulation environment and present simulation results that support the gained insights regarding context prediction.

We study the impact of the context interpretation error on the context prediction accuracy. Benefits and drawbacks of current context prediction schemes are analyzed and opposed to a contemporary alternative. We propose a novel context prediction scheme that has the potential to significantly improve the context prediction accuracy. The impact of the context interpretation error on the context prediction accuracy is further analyzed in simulations inspired by our analytical considerations

The ability to predict future contexts significantly expands the possibilities of context-aware computing applications. However, an incorrect prediction may also mislead the application and may result in inappropriate application behaviour. We study influences on the prediction accuracy and propose a novel approach to context prediction in ubiquitous computing environments. In our paper we introduce a context time series prediction algorithm based on local alignment techniques. Our approach has the potential to improve the prediction accuracy since it explores the observed context history in more detail than current algorithms. In conclusion, we present simulation results that support our studies