The article titled Improved RSSI Indoor Localization in IoT Systems with Machine Learning Algorithms by Ruvan Abeysekera and Ruvan Abeysekera focuses on enhancing indoor localisation in Internet of Things (IoT) systems using AI machine learning algorithms. The paper addresses the limitations of GPS in indoor environments and explores the use of Bluetooth low-energy (BLE) nodes and Received Signal Strength Indicator (RSSI) values for more accurate localisation.
GPS is ineffective indoors so the paper emphasises the need for alternative methods for indoor localisation, which is crucial for various applications like smart cities, transportation and emergency services.
The study uses machine learning algorithms to process RSSI data collected from Bluetooth nodes in complex indoor environments. Algorithms like K-Nearest Neighbors (KNN), Support Vector Machine (SVM, and Feed Forward Neural Networks (FFNN) are used, achieving accuracies of approximately 85%, 84%, and 76% respectively.
The RSSI data is also processed using techniques like weighted least-squares method and moving average filters. The paper also discusses the importance of hyperparameter tuning in improving the performance of the machine learning models.
The research claims to provide significant advancement in indoor localisation, highlighting the potential of machine learning in overcoming the limitations of traditional GPS-based systems in indoor environments.
This system integrates a ‘blind’ device and multiple base stations using iBeacon components to form virtual digital electronic fences, effectively receiving signals from moving devices or tags in a targeted area. The proposed method divides the positioning area into rectangular or triangular subareas and establishes a loss value database for improved location estimation.
The system shows high accuracy, with an average error of less than 0.5 m in the worst-case scenario, making it suitable for various environments. The paper covers the architecture of the system, development phases and experimental results demonstrating the system’s effectiveness. The research offers significant insights into low-cost, high-precision indoor positioning methods suitable for diverse applications such as healthcare, smart cities, and industrial settings.
Recent research A Novel Optimized iBeacon Localization Algorithm Modeling by Jiajia Shi et al, addresses the challenges in achieving high accuracy in indoor object localisation or tracking using iBeacon systems. These systems, which use Bluetooth sensors, are appealing due to their low cost and ease of setup but there can be challenges with accuracy and they can sometimes be susceptible to interference and environmental noise.
To overcome these challenges, the study focuses on developing error modeling algorithms for signal calibration, uncertainty reduction and noise elimination. The novel approach is based on the Curve Fitted Kalman Filter (CFKF) algorithms. The research demonstrates that the CFKF algorithms significantly improve the accuracy and precision of iBeacon localisation.
The paper discusses the limitations of current indoor localisation technologies, including the Received Signal Strength Indicator (RSSI) method, which is affected by multipath fading in indoor environments.
The authors propose a novel CFKF error modelling approach to enhance the estimation accuracy of iBeacon systems in field experiments. This approach includes a developed Kalman Filter (KF) state estimate algorithm based on the modified Least Squares Algorithm (LSA), a system calibration process for the RSSI and estimated distance and the CFKF error modelling for improved accuracy.
The physical system consists of an observation (they call blind) device and multiple base stations using iBeacon components. These base stations can form virtual digital electronic fences and receive signals from blind devices, such as wearable devices or equipment tags. The positioning area is divided into rectangular or triangular subareas and the location of a blind device can be accurately located in real time using the measured strength of received signals and topology analysis.
The proposed method has an average error of less than 0.5 meters in the worst scenario and can be easily used in various environments. It integrates an STSS database and a triangulation method by evaluating the power values of received directional signals. Compared to traditional triangulation technologies, this method offers better positioning accuracy with simpler implementation procedures, reducing the overall cost of deployment.
Historically, many fingerprinting models that harness received signal strength (RSS) data have been proposed for indoor positioning. However, a large number of these methods require an extensive installation of beacons. In contrast, proximity estimation, which relies solely on a single beacon, emerges as a more apt solution, especially for vast automated laboratories.
The intricacies of the life science automation laboratory environment present hurdles for the conventional path loss model (PLM), a prevalent method of proximity estimation based on radio wave propagation. Addressing this challenge, the paper introduces BLE sensing devices crafted specifically for sample position estimation. The proximity estimation rooted in BLE beacon technology is explored within a machine learning framework. Here, support vector regression (SVR) is employed to capture the nonlinear correlation between RSS data and distance. Concurrently, the Kalman filter is applied to reduce deviations in the RSS data.
Experimental outcomes spanning diverse settings underline the superiority of SVR over PLM. Remarkably, SVR achieved 1m absolute errors for an impressive 95% of test samples. The addition of the Kalman filter augments stable distance predictions, effectively smoothed the raw data and mitigated extreme value impacts.
When estimating positions between parallel workbenches, the framework achieved an average mean absolute error (MAE) of just 0.752m across 12 test positions. And for position estimation on workstations, identification accuracies beyond 99.93%.
In conclusion, for labs aiming to enhance sample position estimation, the BLE beacon paired with an IoT node presents a flexible sensing solution. By integrating machine learning, particularly SVR, and the Kalman filter, this framework offers increased accuracy in both corridors and labs.
Most motion sensing applications usually place a sensor beacon on the things that will move. The accelerometer in the beacon reports movement. The research paper describes an alternative method of detecting movement of a person, an elderly person in this case, based on the change in blocking of the beacon signal over time. This has the advantage that the beacon doesn’t need to be worn. Also, it doesn’t have to be a accelerometer beacon as any beacon can be used.
The problem with using the strength of the beacon signal (RSSI), is that it varies over time even when there’s no change of blocking in the room. This is due to radio frequency (RF) noise and reflection. The authors of the paper looked into smoothing of the data to filter out such variance in the data:
The report concludes that when averaging over three or more RSSI values, it’s possible to minimise the RF variance and reliably detect the variance caused by human movement in the room.
Another, more reliable, way of detecting movement is to use a beacon with built-in PIR such as the iBS02PIR, M52-PIR, IX32 or MSP01.
There’s new research from the University of Illinois titled Packet Reception Probability: Packets That You Can’t Decode Can Help Keep You Safe (pdf). Many existing systems estimate distance using the Receiver Signal Strength Indicator (RSSI) which is negatively impacted by sampling bias and multipath effects. As an alternative, the study uses Packet Reception Probability (PRP) that utilises packet loss to estimate distance.
Localisation is achieved through a Bayesian-PRP approach that also includes an explicit model of multipath. To facilitate straightforward deployment, there’s no need for any modifications to hardware, firmware, or driver-level on standard devices and only minimal training is required.
A variety of devices were used including Bluvision iBeeks, BluFi, a Texas Instrument Packet Sniffer, a laptop, and Android smartphones (Nexus5x). 60 iBeacons were deployed in a library and 38 in a retail store. The Texas Instrument Packet Sniffer, connected to a Windows laptop was used for packet reception from beacons. Android phones were equipped with a purpose-built Android app.
PRP was found to provide metre-level accuracy with just six devices in known locations and 12 training locations. Combining PRP with RSSI was found to be beneficial at short distances up to 2m. Beyond distances of 2m, fusing the two is less effective than using PRP alone because RSSI becomes de-correlated with distance.
Bluetooth SIG, the organisation responsible for Bluetooth standards, has a new Bluetooth® Market Update in collaboration with ABI Research. Bluetooth covers a large range of device types and application areas. Here are some insights related to location services.
Bluetooth location services device growth will trend significantly upward and return to pre-pandemic forecasts due to heightened awareness of the benefits of Bluetooth location services. There will be 2.46x growth in annual Bluetooth location services device shipments from 2023 to 2027.
Bluetooth real time location systems (RTLS) are set for rapid growth. New regulatory and safety requirements in manufacturing, stricter compliance procedures and sustainable operation requirements are making RTLS solutions more attractive. There will be 178,000 Bluetooth® RTLS implementations by the end of 2023. Many commercial and industrial facilities are now relying on asset tracking solutions to optimise resource and inventory control. The commoditisation of off-the-shelf Bluetooth asset tracking gateways and beacons are major drivers behind continued growth. 112 million Bluetooth asset tracking devices will ship in 2023.
A smartphone app was used to receive beacon RSSI and send it to an indoor positioning system’s data collection module. RSSI data was also filtered by a data processing module to limit the error range. KNN, RFC, extra trees classifiers (ETC), SVM, gradient boosting classifiers (GBC) and decision trees (DT) algorithms were evaluated.
The ETC model gave the best accuracy. ETC is an algorithm that uses a group of decision trees to classify data. It is similar to a random forest classifier but uses a different method to construct the decision trees. ETC fits a number of randomised decision trees on sub-samples of the dataset and uses averaging to improve the predictive accuracy and control over-fitting. ETC is a good choice for applications where accuracy is important but the data is noisy and where computational efficiency is important.
Most Beacon usecases involve putting beacons on things or in places and triggering notifications on users’ phones. There’s a paper by Yang Yang, Zhouchi Li and Kaveh Pahlavan of Worcester Polytechnic Institute (WPI), Worcester, MA that instead proposes Using iBeacon for Intelligent In-Room Presence Detection.
Their system records users in a room for applications such as graduate seminar check-in, security and in and out counting. It recognises in room presence by analysing path loss and door motion readings to decide whether a person is inside the room. Their custom app receives the beacon data and sends it to a server for analysis. They experimented using two iBeacons, one attached to the outside of the door with another mirroring at the inside and also as single iBeacon implementation that still performed well.
The paper also a useful chart showing the variation of RSSI with how a phone is held: