With the rapid development of smart cities, smart buildings are generating a massive amount of building sensing data by the equipped sensors. Indeed, building sensing data provides a promising way to enrich a series of data-demanding and cost-expensive urban mobile applications. In this paper, as a preliminary exploration, we study how to reuse building sensing data to predict traffic volume on nearby roads. Compared with existing studies, reusing building sensing data has considerable merits of cost-efficiency and high-reliability. Nevertheless, it is non-trivial to achieve accurate prediction on such cross-domain data with two major challenges. First, relationships between building sensing data and traffic data are not unknown as prior, and the spatio-temporal complexities impose more difficulties to uncover the underlying reasons behind the above relationships. Second, it is even more daunting to accurately predict traffic volume with dynamic building-traffic correlations, which are cross-domain, non-linear, and time-varying. To address the above challenges, we design and implement BuildSenSys , a first-of-its-kind system for nearby traffic volume prediction by reusing building sensing data. Our work consists of two parts, i.e., Correlation Analysis and Cross-domain Learning . First, we conduct a comprehensive building-traffic analysis based on multi-source datasets, disclosing how and why building sensing data is correlated with nearby traffic volume. Second, we propose a novel recurrent neural network for traffic volume prediction based on cross-domain learning with two attention mechanisms. Specifically, a cross-domain attention mechanism captures the building-traffic correlations and adaptively extracts the most relevant building sensing data at each predicting step. Then, a temporal attention mechanism is employed to model the temporal dependencies of data across historical time intervals. The extensive experimental studies demonstrate that BuildSenSys outperforms all baseline methods with up to 65.3 percent accuracy improvement (e.g., 2.2 percent MAPE) in predicting nearby traffic volume. We believe that this work can open a new gate of reusing building sensing data for urban traffic sensing, thus establishing connections between smart buildings and intelligent transportation.

Indoor localization has become an essential demand driven by indoor location-based services (ILBSs) for mobile users. With the rising of Internet of Things (IoT), heterogeneous smartphones and wearables have become ubiquitous. However, the ILBSs for heterogeneous IoT devices confront significant challenges, such as received signal strength (RSS) variances caused by hardware heterogeneity, multipath reflections from complex environments, and localization time restricted by computation resources. This article proposes EdgeLoc, a robust and real-time indoor localization system toward heterogeneous IoT devices to solve the above challenges. In particular, the RSS fingerprinting data of Wi-Fi is employed for localization and tackling the heterogeneity of IoT devices in twofold. First, feature-level and signal-level solutions are presented to address the random RSS variances. At the feature level, this work proposes a novel capsule neural network model to efficiently extract incremental features from RSS fingerprinting data. At the signal level, a multistep dataflow is further devised to process RSS fingerprints into image-like data, which utilizes the feature matrix to reduce absolute sensing errors introduced by hardware heterogeneity. Second, an edge-IoT framework is designed to utilize the edge server to train the deep learning model and further supports real-time localization for heterogeneous IoT devices. Extensive field experiments with over 33 600 data points are conducted to validate the effectiveness of EdgeLoc with a large-scale Wi-Fi fingerprint data set. The results show that EdgeLoc outperforms the state-of-the-art SAE-CNN method in localization accuracy by up to 14.4%, with an average error of 0.68 m and an average positioning time of 2.05 ms.

WiFi infrastructures are widely deployed in both public and private buildings. They make the connection to the internet more convenient. Recently, researchers find that WiFi signals have the ability to sense the changes in the environment that can detect human motion and even identify human activities and his identity in a device-free manner, and has many potential security applications in a smart home. Previous human detection systems can only detect human motion of regular moving patterns. However, they may have a significant detection performance degradation when used in intrusion detection. In this study, we propose Robust Device-Free Intrusion Detection (RDFID) system leveraging fine-grained Channel State Information (CSI). The noises in the signals are removed by a Principle Component Analysis (PCA) and a low pass filter. We extract a robust feature of frequency domain utilizing Continuous Wavelet Transform (CWT) from all subcarriers. RDFID captures the changes from the whole wireless channel, and a threshold is obtained self-adaptively, which is calibration-free in different environments, and can be deployed in smart home scenarios. We implement RDFID using commodity WiFi devices and evaluate it in three typical office rooms with different moving patterns. The results show that our system can accurately detect intrusion of different moving patterns and different environments without re-calibration.

Doors as densely-deployed natural landmarks play an important role in improving indoor positioning systems. However, the state-of-the-art door event detection works are based on either vision or infrastructure, thus incurring non-trivial device or management cost. To address these problems, we present a Light-weight Magnetic-based Door Event Detection method, called LMDD. It leverages built-in magnetic sensors of common smartphones to achieve infrastructure-free door event detection. After analyzing the special features of sensors' readings changes caused by the door, we design LMDD scheme with three main components, including data acquisition, events identification and events denoising. Moreover, an improved and robust door event detection framework based on a majority-voting model is proposed to fuse multiple-dimensional sensing data from non-magnetic built-in sensors. We have implemented a prototype of LMDD on Android-based platform. Experimental results show that LMDD with only magnetic sensor achieves door event detection accuracy of around 80% on average, ranging from 70% to 87% in various typical indoor environments. The enhanced LMDD based on the fusion of heterogeneous sensors can achieve a much higher door event detection accuracy of 90% on average.

Android applications have emerged as a prime target for hackers. Android malware detection stands as a pivotal technology, crucial for safeguarding network security and thwarting anomalies. However, traditional static analysis makes it difficult to analyze new malicious applications, while dynamic analysis requires higher system resources. We propose a novel lightweight Android malware deep-learning detection framework based on attention temporal networks. This study delves into the Dalvik opcode sequences of Android malware, employing the N-gram algorithm to partition sequences and extract contextual information features. Then, LSTM and TCN algorithms are employed to capture long-term dependencies and local features, enabling comprehensive comprehension of temporal information within Dalvik opcode sequences. Especially, TCN facilitates feature extraction across various time scales, thereby enabling the detection of anomaly patterns across diverse temporal scales within Dalvik opcode sequences. Moreover, we introduce multi-head attention mechanisms and reinforced learning to direct the model’s focus toward behavioral cues within malicious software sequences. Finally, extensive experiment results show that our proposed methodology and model exhibit higher detection accuracy and robustness, achieving an accuracy rate of 98.69% on average, surpassing traditional machine learning methods such as random forest, and Pseudo-Label deep neural networks.