No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Wi-Fi positioning based on deep learning
... In this study, we take a fingerprint-based approach [5,6,7,8,9] using a database that stores the relationship between RSSI and position as an existing study of localization using low power wireless systems. This approach can be divided into "matching approach" and "deep learning approach." ...
... This approach can be divided into "matching approach" and "deep learning approach." The matching approach is a method using pattern matching between RSSI at the time of localization and RSSI in the database [5,6] and the deep learning approach is a method using the database as learning data in deep learning [7,8,9]. However, to realize accurate estimation over a vast outdoor environment inevitably leads to the large size of the fingerprint database, so that the matching approach will suffer from a long retrieval time in the database. ...
We are developing a method to acquire position information of a cow outdoors using Received Signal Strength Indicator (RSSI) of Bluetooth Low Energy (BLE). As existing research, there is a localization method using fingerprint database as learning data in deep learning. However, that method has the problem that it costs to create a database by measurement in a vast outdoor environment. Therefore, we considered to build a part of the fingerprint database using virtual space modeling received power measurement environment in a pasture. Experimental results showed that an average distance error to GPS data is about 6 m by training DNN using the database and additionally training DNN using actual GPS data.
... CNNs have achieved impressive performance in various computer vision tasks [9], [10], [11], [12], [13], [14], yet so far using CNNs for room layout estimation is rare. To estimate room layouts with CNNs, the primary problem is how to denote the room layouts appropriately as the network output. ...
The goal of room layout estimation is to predict the 3D box that represents the room spatial structure from a monocular image. In this paper, a deconvolution network is trained firstly to predict the edge map of a room image. Compared to the previous fully convolutional networks, the proposed deconvolution network has multi-layer deconvolution process which can refine the edge map estimate layer by layer. The deconvolution network also has fully connected layers to aggregate the information of every region throughout the entire image. During the layout generation process, an adaptive sampling strategy is introduced based on the obtained high-quality edge maps. Experimental results prove that the learned edge maps are highly reliable and can produce accurate layouts of room images.
... To test the scalability of the proposed method, the total 100N fingerprints are randomly sampled to form additional four sample sets with different sizes such as 20N, 40N, 60N and 80N. We tested the performance of the proposed method in terms of root mean square error (RMSE) in meters with different sizes of sample sets, and compared with some existing methods such as KNN [3], SVM [4], Locally Linear Embedding (LLE) [38], RBMþBP [39], as well as their combinations with HMM. As shown in Table 1, the proposed deep network generally maintains good performance for all conditions as the sample size grows. ...
In this paper, we propose a wireless positioning method based on Deep Learning. To deal with the variant and unpredictable wireless signals, the positioning is casted in a four-layer Deep Neural Network (DNN) structure pre-trained by Stacked Denoising Autoencoder (SDA) that is capable of learning reliable features from a large set of noisy samples and avoids hand-engineering. Also, to maintain the temporal coherence, a Hidden Markov Model (HMM)-based fine localizer is introduced to smooth the initial positioning estimate obtained by the DNN-based coarse localizer. The data required for the experiments is collected from the real world in different periods to meet the actual environment. Experimental results indicate that the proposed system leads to substantial improvement on localization accuracy in coping with the turbulent wireless signals.
Recently, various application fields utilizing Wi-Fi fingerprint data have been under research. However, fingerprint data collected from a specific location does not include relevant information, such as continuity. Therefore, most indoor positioning technologies predict the user’s location based on location signals collected at specific points within the indoor space, without taking into account the user’s movements. Hence, there is a need for technology that improves the accuracy of indoor positioning while moving. This paper proposes a technique to generate movement path data by applying the concepts of “BB” and “Grid Cell” from computer vision to Wi-Fi fingerprint data. This approach represents data points as bounding boxes (BBs), then establishes grid cells and clusters of these BBs to generate adjacency information. Subsequently, movement path data are created based on this information. We utilized the movement path information generated from the dataset as training data for machine learning and introduced an enhanced indoor positioning technology. First, the experiments in this paper assessed the performance of the proposed technology based on the number of paths in the LSTM model. Second, the performance of clustering methods was compared through experiments. Finally, we evaluated the performance of various machine learning models. The experimental results confirmed a maximum accuracy of 94.48% when determining the location based on route information. Clustering performance improved accuracy by up to 3%. In comparative experiments with machine learning models, accuracy improved by up to 2.8%.
Control measures have been applied in recent years due to the COVID-19 pandemic. Different technologies including artificial intelligence (AI) and geofencing are required to be exploited for developing efficient techniques to deal with this crisis. Workplaces are the most dangerous areas that can lead to the infection of the pandemic. This is due to the increased density of people and transactions in limited places. In this paper, an efficient approach is proposed to monitor and impose COVID-19 control measures in workplaces. The workplace environment is clustered based on a dynamic user-centric clustering scheme, where each person in the workplace is assigned to a set of associated geofences that form its cluster. For each geofence, different wireless and network metrics are used for generating its digital signature. An efficient technique based on deep learning is proposed to generate the geofence digital signature and detect whether the person is inside his associated cluster or not. Experimental results show the effectiveness of the proposed technique for different locations in a real workplace.
We present FALCON, a novel autonomous drone network system for sensing, localizing, and approaching RF targets/sources such as smartphone devices. Potential applications of our system include disaster relief missions in which networked drones sense the Wi-Fi signal emitted from a victim’s smartphone and dynamically navigate to accurately localize and quickly approach the victim, for instance, to deliver the time-critical first-aid kits. For that we exploit Wi-Fi’s recent fine time measurement (FTM) protocol to realize the first on-drone FTM sensor network that enables accurate and dynamic ranging of targets in a mission. We propose a flight planning strategy that adapts the trajectory of the drones to concurrently favor localizing and approaching the target. Namely, our approach jointly optimizes the drones’ diversity of observations and the target approaching process, while flexibly trading off the intensities of the potentially conflicting objectives. We implement FALCON via a custom-designed multidrone platform and demonstrate up to
localization accuracy compared to a baseline flocking approach, while spending 30% less time localizing targets.
One of the most popular approaches for indoor positioning is WiFi fingerprinting, which has been intrinsically tackled as a traditional machine learning problem since the beginning, to achieve a few metres of accuracy on average. In recent years, deep learning has emerged as an alternative approach, with a large number of publications reporting sub-metre positioning accuracy. Therefore, this survey presents a timely, comprehensive review of the most interesting deep learning methods being used for WiFi fingerprinting. In doing so, we aim to identify the most efficient neural networks, under a variety of positioning evaluation metrics for different readers. We will demonstrate that despite the new emerging WiFi signal measures (i.e. CSI and RTT), RSS produces competitive performances under deep learning. We will also show that simple neural networks outperform more complex ones in certain environments.
The recurrent neural network (RNN) model, which is a deep-learning network that can memorize past information, is used in this paper to memorize continuous movements in indoor positioning to reduce positioning error. To use an RNN model in Wi-Fi-fingerprint based indoor positioning, data set must be sequential. However, Wi-Fi fingerprinting only saves the received signal strength indicator for a location, so it cannot be used as RNN data. For this reason, we propose a movement path data generation technique that generates data for an RNN model for sequential positioning from Wi-Fi fingerprint data. Movement path data can be generated by creating an adjacency list for Wi-Fi fingerprint location points. However, creating an adjacency matrix for all location points requires a large amount of computation. This problem is solved by dividing indoor environment by K-means clustering and creating a cluster transition matrix based on the center of each cluster.
The problem of position estimation has always been widely discussed in the field of wireless communication. In recent years, deep learning technology is rapidly developing and attracting numerous applications. The high-dimension modeling capability of deep learning makes it possible to solve the localization problems under many nonideal scenarios which are hard to handle by classical models. Consequently, wireless localization based on deep learning has attracted extensive research during the last decade. The research and applications on wireless localization technology based on deep learning are reviewed in this paper. Typical deep learning models are summarized with emphasis on their inputs, outputs, and localization methods. Technical details helpful for enhancing localization ability are also mentioned. Finally, some problems worth further research are discussed.
This contribution describes a Fingerprint position estimation using RSSI (Received Signal Strength Indicator) of wireless LAN Access Points (APs) that employs Neural Network (NN). We propose position estimation using the weighted intermediate point to estimate the exact position. In this method, RSSI of UD (User Data) is collated with RSSI of DB (Data Base). RSSI of DB was measured on coordinates in advance. Therefore, one of the coordinates is selected as the estimation result. The coordinate closest to correct point is selected in case of estimation at the intermediate point of the coordinates. In this study, we propose to estimate the intermediate point using weighting with the existence probability distribution of limited coordinate output from NN. The accuracy of proposed position estimation method was verified using three-layer NN based on measured data.
With the superior capability of discovering intricate structure of large data sets, deep learning has been widely applied in various areas including wireless networking. While existing deep learning applications mainly focus on data analysis, the role it can play on fundamental research issues in wireless networks is yet to be explored. With the proliferation of wireless networking infrastructure and mobile applications, wireless network optimization has seen a tremendous increase in problem size and complexity, calling for a paradigm for efficient computation. This paper presents a pioneering study on how to exploit deep learning for significant performance gain in wireless network optimization. Analysis on the flow constrained optimization problems suggests the possibility that a smaller-sized problem can be solved while sharing equally optimal solutions with the original problem, by excluding the potentially unused links from the problem formulation. To this end, we design a deep learning framework to find the latent relationship between flow information and link usage by learning from past computation experience. Numerical results demonstrate that the proposed method is capable of identifying critical links and can reduce computation cost by up to 50% without affecting optimality, thus greatly improve the efficiency of solving network optimization problems.
In this paper, we propose to learn the structures of stereoscopic image based on convolutional neural network (CNN) for no-reference quality assessment. Taking image patches from the stereoscopic images as inputs, the proposed CNN can learn the local structures which are sensitive to human perception and representative for perceptual quality evaluation. By stacking multiple convolution and max-pooling layers together, the learned structures in lower convolution layers can be composed and convolved to higher levels to form a fixed-length representation. Multilayer perceptron (MLP) is further employed to summarize the learned representation to a final value to indicate the perceptual quality of the stereo image patch pair. With different inputs, two different CNNs are designed, namely one-column CNN with only the image patch from the difference image as input, and three-column CNN with the image patches from left-view image, right-view image, and difference image as the input. The CNN parameters for stereoscopic images are learned and transferred based on the large number of 2D natural images. With the evaluation on public LIVE phase-I, LIVE phase-II, and IVC stereoscopic image databases, the proposed no-reference metric achieves the state-of-the-art performance for quality assessment of stereoscopic images, and is even competitive to existing full-reference quality metrics.
The recent growing interest for indoor Location-Based Services (LBSs) has created a need for more accurate and real-time indoor positioning solutions. The sparse nature of location finding makes the theory of Compressive Sensing (CS) desirable for accurate indoor positioning using Received Signal Strength (RSS) from Wireless Local Area Network (WLAN) Access Points (APs). We propose an accurate RSS-based indoor positioning system using the theory of compressive sensing, which is a method to recover sparse signals from a small number of noisy measurements by solving an `1-minimization problem. Our location estimator consists of a coarse localizer, where the RSS is compared to a number of clusters to detect in which cluster the node is located, followed by a fine localization step, using the theory of compressive sensing, to further refine the location estimation. We have investigated different coarse localization schemes and AP selection approaches to increase the accuracy. We also show that the CS theory can be used to reconstruct the RSS radio map from measurements at only a small number of fingerprints, reducing the number of measurements significantly. We have implemented the proposed system on a WiFi-integrated mobile device and have evaluated the performance. Experimental results indicate that the proposed system leads to substantial improvement on localization accuracy and complexity over the widely used traditional fingerprinting methods.
This paper exploits recent developments in sparse approximation and compressive sensing to efficiently perform localization in wireless networks. Particularly, we re-formulate the localization problem as a sparse approximation problem using the compressive sensing theory that provides a new paradigm for recovering a sparse signal solving an ℓ1 minimization problem. The proposed received signal strength-based method does not require any time specific/propriatery hardware since the location estimation is performed at the Access Points (APs). The experimental results show that our proposed method, when compared with traditional localization schemes results in a better accuracy in terms of the mean localization error.
Gaussian mixture models are currently the dominant technique for modeling the emission distribution of hidden Markov models for speech recognition. We show that better phone recognition on the TIMIT dataset can be achieved by replacing Gaussian mixture models by deep neural networks that contain many layers of features and a very large number of parameters. These networks are first pre-trained as a multi-layer generative model of a window of spectral feature vectors without making use of any discriminative information. Once the generative pre-training has designed the features, we perform discriminative fine-tuning using backpropagation to adjust the features slightly to make them better at predicting a probability distribution over the states of monophone hidden Markov models.
In recent years, deep learning approaches have gained significant interest as a way of building hierarchical representations from unlabeled data. However, to our knowledge, these deep learning approaches have not been extensively stud- ied for auditory data. In this paper, we apply convolutional deep belief net- works to audio data and empirically evaluate them on various audio classification tasks. In the case of speech data, we show that the learned features correspond to phones/phonemes. In addition, our feature representations learned from unlabeled audio data show very good performance for multiple audio classification tasks. We hope that this paper will inspire more research on deep learning approaches applied to a wide range of audio recognition tasks.
Authors presented recently an indoor location technique based on Time Of Arrival (TOA) obtained from Round-Trip-Time (RTT) measurements at data link level and trilateration. This new approach uses the existing IEEE 802.11 WLAN infrastructure with minor changes to provide an accurate estimation of the position of static wireless terminals. This paper presents advances on how to incorporate tracking capabilities to this approach in order to achieve a noticeable enhancement in the positioning accuracy while maintaining the computational cost low, both essential requirements in some critical applications of indoor pedestrian navigation in which people carrying light mobile devices has to be tracked with precision. Taking as a basis the Discrete Kalman Filter, customizations and optimizations have been designed and presented. Results obtained after conducting extensive simulations fed with actual ranging observables demonstrate the validity and suitability of the researched algorithms and its ability to provide very high performance level in terms of accuracy and robustness. Peer reviewed
It is possible to combine multiple latent-variable models of the same data by multiplying their probability distributions together and then renormalizing. This way of combining individual “expert” models makes it hard to generate samples from the combined model but easy to infer the values of the latent variables of each expert, because the combination rule ensures that the latent variables of different experts are conditionally independent when given the data. A product of experts (PoE) is therefore an interesting candidate for a perceptual system in which rapid inference is vital and generation is unnecessary. Training a PoE by maximizing the likelihood of the data is difficult because it is hard even to approximate the derivatives of the renormalization term in the combination rule. Fortunately, a PoE can be trained using a different objective function called “contrastive divergence” whose derivatives with regard to the parameters can be approximated accurately and efficiently. Examples are presented of contrastive divergence learning using several types of expert on several types of data.
We show how to use “complementary priors” to eliminate the explaining-away effects that make inference difficult in densely connected belief nets that have many hidden layers. Using complementary priors, we derive a fast, greedy algorithm that can learn deep, directed belief networks one layer at a time, provided the top two layers form an undirected associative memory. The fast, greedy algorithm is used to initialize a slower learning procedure that fine-tunes the weights using a contrastive version of the wake-sleep algorithm. After fine-tuning, a network with three hidden layers forms a very good generative model of the joint distribution of handwritten digit images and their labels. This generative model gives better digit classification than the best discriminative learning algorithms. The low-dimensional manifolds on which the digits lie are modeled by long ravines in the free-energy landscape of the top-level associative memory, and it is easy to explore these ravines by using the directed connections to display what the associative memory has in mind.
Location based services (LBS), context aware applications, and people and object tracking depend on the ability to locate mobile devices, also known as localization, in the wireless landscape. Localization enables a diverse set of applications that include, but are not limited to, vehicle guidance in an industrial environment, security monitoring, self-guided tours, personalized communications services, resource tracking, mobile commerce services, guiding emergency workers during fire emergencies, habitat monitoring, environmental surveillance, and receiving alerts. This paper presents a new neural network approach (LENSR) based on a competitive topological counter propagation network (CPN) with k-nearest neighborhood vector mapping, for indoor location estimation based on received signal strength. The advantage of this approach is both speed and accuracy. The tested accuracy of the algorithm was 90.6% within 1 meter and 96.4% within 1.5 meters. Several approaches for location estimation using WLAN technology were reviewed for comparison of results.
Context is a critical ingredient of ubiquitous computing. While it is possible to use specialized sensors and beacons to measure certain aspects of a user's context, we are interested in what we can infer from using the existing 802.11 wireless network infrastructure that already exists in many places. The context parameters we infer are the location of a client (with a median error of 1.5 meters) and an indicator of whether or not the client is in motion (with a classification accuracy of 87%). Our system, called LOCADIO, uses Wi-Fi signal strengths from existing access points measured on the client to infer both pieces of context. For motion, we measure the variance of the signal strength of the strongest access point as input to a simple two-state hidden Markov model (HMM) for smoothing transitions between the inferred states of "still" and "moving". For location, we exploit the fact that Wi-Fi signal strengths vary with location, and we use another HMM on a graph of location nodes whose transition probabilities are a function of the building's floor plan, expected pedestrian speeds, and our still/moving inference. Our probabilistic approach to inferring context gives a convenient way of balancing noisy measured data such as signal strengths against our a priori assumptions about a user's behavior.
We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 dif-ferent classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make train-ing faster, we used non-saturating neurons and a very efficient GPU implemen-tation of the convolution operation. To reduce overfitting in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.
Fingerprinting localization techniques have been intensively studied in indoor WLAN environment. Artificial neural networks (ANN) based fingerprinting technique could potentially provide high accuracy and robust performance. However, it has the limitations of slow convergence, high complexity and large memory storage requirement, which are the bottlenecks of its wide application, especially in the case of a large-scale indoor environment and the terminal with limited computing capability and memory resources. In this paper, we firstly introduce affinity propagation (AP) clustering algorithm to reduce the computation cost and memory overhead, and then explore the properties of radio basis function (RBF) neural networks that may affect the accuracy of the proposed fingerprinting localization systems. We carry out various experiments in a real-world setup where multiple access points are present. The detailed comparison results reveal how the clustering algorithm and the neural networks affect the performance of the proposed algorithms.
We consider the problem of building high-level, class-specific feature detectors from only unlabeled data. For example, is it possible to learn a face detector using only unlabeled images? To answer this, we train a deep sparse autoencoder on a large dataset of images (the model has 1 billion connections, the dataset has 10 million 200×200 pixel images downloaded from the Internet). We train this network using model parallelism and asynchronous SGD on a cluster with 1,000 machines (16,000 cores) for three days. Contrary to what appears to be a widely-held intuition, our experimental results reveal that it is possible to train a face detector without having to label images as containing a face or not. Control experiments show that this feature detector is robust not only to translation but also to scaling and out-of-plane rotation. We also find that the same network is sensitive to other high-level concepts such as cat faces and human bodies. Starting from these learned features, we trained our network to recognize 22,000 object categories from ImageNet and achieve a leap of 70% relative improvement over the previous state-of-the-art.
The proliferation of mobile computing devices and local-area wireless networks has fostered a growing interest in location-aware systems and services. In this paper we present RADAR, a radio-frequency (RF) based system for locating and tracking users inside buildings. RADAR operates by recording and processing signal strength information at multiple base stations positioned to provide overlapping coverage in the area of interest. It combines empirical measurements with signal propagation modeling to determine user location and thereby enable location-aware services and applications. We present experimental results that demonstrate the ability of RADAR to estimate user location with a high degree of accuracy.
We propose a novel context-dependent (CD) model for large-vocabulary speech recognition (LVSR) that leverages recent advances in using deep belief networks for phone recognition. We describe a pre-trained deep neural network hidden Markov model (DNN-HMM) hybrid architecture that trains the DNN to produce a distribution over senones (tied triphone states) as its output. The deep belief network pre-training algorithm is a robust and often helpful way to initialize deep neural networks generatively that can aid in optimization and reduce generalization error. We illustrate the key components of our model, describe the procedure for applying CD-DNN-HMMs to LVSR, and analyze the effects of various modeling choices on performance. Experiments on a challenging business search dataset demonstrate that CD-DNN-HMMs can significantly outperform the conventional context-dependent Gaussian mixture model (GMM)-HMMs, with an absolute sentence accuracy improvement of 5.8% and 9.2% (or relative error reduction of 16.0% and 23.2%) over the CD-GMM-HMMs trained using the minimum phone error rate (MPE) and maximum-likelihood (ML) criteria, respectively.
Indoor localisation technologies have received considerable attention in recent years and they are applicable in many fields. As with outdoor systems, they both suffer a degradation in performance when multi-path errors are present. Applications include locating essential equipment in hospitals and specific items in warehouses, tracking people with special needs, who are away from visual supervision, and navigating fire-fighters inside buildings. In this article, we study issues associated with the implementation of a real-time Wireless Fidelity (WiFi) localisation system for an indoor environment. In particular, we present robust techniques to mitigate the effects of the multi-path errors. This system utilises smart antennas to determine the signal strength information from a mobile target (MT) (access point) and send the information to a data processing station. This information is combined to find the direction of arrival of the signal and triangulate the MT position. No prior finger-printing is required, as this system avoids the use of an off-line training phase, which is computationally intensive and requires a big database. This approach is more computationally efficient and non-data intensive. Experimental results show an improvement in the accuracy of the localisation system over conventional techniques.
Many areas of science depend on exploratory data analysis and visualization. The need to analyze large amounts of multivariate
data raises the fundamental problem of dimensionality reduction: how to discover compact representations of high-dimensional
data. Here, we introduce locally linear embedding (LLE), an unsupervised learning algorithm that computes low-dimensional,
neighborhood-preserving embeddings of high-dimensional inputs. Unlike clustering methods for local dimensionality reduction,
LLE maps its inputs into a single global coordinate system of lower dimensionality, and its optimizations do not involve local
minima. By exploiting the local symmetries of linear reconstructions, LLE is able to learn the global structure of nonlinear
manifolds, such as those generated by images of faces or documents of text.
The proliferation of mobile computing devices and local-area
wireless networks has fostered a growing interest in location-aware
systems and services. In this paper we present RADAR, a radio-frequency
(RF)-based system for locating and tracking users inside buildings.
RADAR operates by recording and processing signal strength information
at multiple base stations positioned to provide overlapping coverage in
the area of interest. It combines empirical measurements with signal
propagation modeling to determine user location and thereby enable
location-aware services and applications. We present experimental
results that demonstrate the ability of RADAR to estimate user location
with a high degree of accuracy