Rafael Berkvens

A realistic evaluation of indoor positioning systems based on Wi-Fi fingerprinting: The 2015 EvAAL–ETRI competition

This paper presents results from comparing different Wi-Fi fingerprinting algorithms on the same private dataset. The algorithms where realized by independent teams in the frame of the off-site track of the EvAAL-ETRI Indoor Localization Competition which was part of the Sixth International Conference on Indoor Positioning and Indoor Navigation (IPIN 2015). Competitors designed and validated their algorithms against the publicly available UJIIndoorLoc database which contains a huge referenceand validation data set. All competing systems were evaluated using the mean error in positioning, with penalties, using a private test dataset. The authors believe that this is the first work in which Wi-Fi fingerprinting algorithm results delivered by several independent and competing teams are fairly compared under the same evaluation conditions. The analysis also comprises a combined approach: Results indicate that the competing systems where complementary, since an ensemble that combines three competing methods reported the overall best results.

DASH7 alliance protocol 1.0: Low-power, mid-range sensor and actuator communication

This paper presents the DASH7 Alliance Protocol 1.0. It is an industry alliance standard for wireless sensor and actuator communication using the unlicensed sub-1 GHz bands. The paper explains its historic relation to active RFID standards ISO 18000-7 for 433 MHz communication, the basic concepts and communication paradigms of the protocol. Since the protocol is a full OSI stack specification, the paper discusses the implementation of every OSI layer.

Localization performance quantification by conditional entropy

The performance of a localization algorithm is usually expressed as its mean error distance. We argue that this assumes a unimodal distribution of the localization posterior, which is not always appropriate. We propose to additionally quantify the localization posterior distribution by its conditional entropy. This informs us of the uncertainty over the position after a measurement, which must be processed by the localization algorithm. Our example measurement model was ranked in the Evaluating Ambient Assisted Living competition, for which we present the results. Furthermore, we discuss the conditional entropy of our measurement model and two additional measurement models, based on the absolute difference distance and the Pompeiu-Hausdorff distance. We compare these results by using the UJIIndoorLoc database that was also used for the competition.

Biologically inspired SLAM using Wi-Fi

Wi-Fi is a commonly available source of localization information in urban environments but is challenging to integrate into conventional mapping architectures. Current state of the art probabilistic Wi-Fi SLAM algorithms are limited by spatial resolution and an inability to remove the accumulation of rotational error, inherent limitations of the Wi-Fi architecture. In this paper we leverage the low quality sensory requirements and coarse metric properties of RatSLAM to localize using Wi-Fi fingerprints. To further improve performance, we present a novel sensor fusion technique that integrates camera and Wi-Fi to improve localization specificity, and use compass sensor data to remove orientation drift. We evaluate the algorithms in diverse real world indoor and outdoor environments, including an office floor, university campus and a visually aliased circular building loop. The algorithms produce topologically correct maps that are superior to those produced using only a single sensor modality.

Conditional Entropy and Location Error in Indoor Localization Using Probabilistic Wi-Fi Fingerprinting.

Localization systems are increasingly valuable, but their location estimates are only useful when the uncertainty of the estimate is known. This uncertainty is currently calculated as the location error given a ground truth, which is then used as a static measure in sometimes very different environments. In contrast, we propose the use of the conditional entropy of a posterior probability distribution as a complementary measure of uncertainty. This measure has the advantage of being dynamic, i.e., it can be calculated during localization based on individual sensor measurements, does not require a ground truth, and can be applied to discrete localization algorithms. Furthermore, for every consistent location estimation algorithm, both the location error and the conditional entropy measures must be related, i.e., a low entropy should always correspond with a small location error, while a high entropy can correspond with either a small or large location error. We validate this relationship experimentally by calculating both measures of uncertainty in three publicly available datasets using probabilistic Wi-Fi fingerprinting with eight different implementations of the sensor model. We show that the discrepancy between these measures, i.e., many location estimates having a high location error while simultaneously having a low conditional entropy, is largest for the least realistic implementations of the probabilistic sensor model. Based on the results presented in this paper, we conclude that conditional entropy, being dynamic, complementary to location error, and applicable to both continuous and discrete localization, provides an important extra means of characterizing a localization method.

Asynchronous, electromagnetic sensor fusion in RatSLAM

We show the predictive value of the mean mutual information rate about the RatSLAM algorithm for electromagnetic sensors and their combinations. We calculated the mean mutual information between positions in the environment and sensor measurements performed at a specific position in the environment and defined the mean mutual information rate depending on the sensors measurement rate. We compare these results to RatSLAM experience maps generated using these sensors and sensor combinations and define a mean error to quantify the spatial quality of the experience map. We conclude that the mean mutual information rate generally predicts the performance correctly, but also find and explain discrepancies in specific cases.

Position error and entropy of probabilistic Wi-Fi fingerprinting in the UJIIndoorLoc dataset

The accuracy of a positioning system is usually expressed as its average position error in an experiment. However, when the ground truth is no longer available, it would still be useful to know the reliability of a position estimate based on a single measurement. To obtain a reliability metric, we hypothesize that there is a relation between the uncertainty in a positions posterior probability distribution, expressed as its conditional entropy, and the position error of the position that is derived from this distribution. In this paper, we present the correlation between these two metrics as calculated for the UJIIndoorLoc Wi-Fi fingerprinting dataset, using a new probabilistic sensor model. We found that there is no significant correlation between the conditional entropy and the position error. However, we learned that our sensor model is usually very certain in the dataset, and saw that the suggestion of a correlation improves when we increase the uncertainty by selecting a fixed, larger variance. Interestingly, the position error results improve as well.

Large Scale Distributed Localization Based on RSS and Mass-Spring Model

Large scale localization has gained interest in the last few years together with wireless sensor networks. Applications such as a swarm of drones or a sensor network to control an environment already exist but do not cover scales up to thousands of nodes. In this paper, a method for large scale distributed localization is developed based on existing technologies including received signal strength and mass-spring model (MSM). First, we created a theoretical model based on specific indoor and outdoor tests. Based on this model, a scaled up simulation was done using an MSM-based algorithm. Using this simulation, an average error of up to 7.66 m could be achieved in a field of 100 m by 100 m.

Contact Detection for Social Networking of Small Animals

Biological research often tracks animal using collars containing a wireless sensor that transmit telemetry or positional data. However, when dealing with small animals, the size and weight of conventional telemetry is often an obstruction and can alter animal behavior. In this study we take a look at the the viability of Bluetooth Low Energy (BLE) to develop a low power contact logger which tracks contacts between small rodents. Using the BLE Discovery Process, a contact logger can reliably detect nearby loggers without the need to set up an actual connection. We manufactured a prototype with an extremely small footprint to demonstrate the feasibility.

Multi-frequency sub-1 GHz radio tomographic imaging in a complex indoor environment

Unlike most currently available localization systems, tagless localization technologies do not require a target to wear a passive or active hardware device. Radio Tomographic Imaging (RTI) is one such technique, which operates based on the use of a tomographic radio frequency (RF) sensor network. The majority of RTI-systems communicate using a single frequency band: 2.4 GHz. The use of sub-1 GHz frequencies within RTI could potentially provide important benefits regarding energy efficiency, accuracy in complex indoor environments and size of the environments in which a system can be installed. We deployed a combined 433 MHz and 868 MHz RF sensor network in a complex indoor environment and performed localization when a human individual was present in the environment. Two different RTI-algorithms were investigated: a Bayesian-based method we developed earlier and an adaptation of an existing 2.4 GHz technique based on fade level. Both methods turned out to be capable of accurately locating individuals with a median error lower than 1 meter. This proves the feasibility of using a combination of sub-1 GHz frequencies in RTI for indoor localization in complex environments.