Citation Context

...ured view to satellites or degraded accuracy caused by multipath. A backup to GPS during signal outage with comparable accuracy could be achieved using fusion of inertial measurement unit (IMU) raw data with already existing cellID based methods and map-matching. The radio profile of a given area can determine routes that are covered by each cell antenna. Therefore, the computational cost of map-matching algorithms would be reduced to a minimum. The proposed positioning algorithm is designed to maintain mobile location information during GPS signal blocking using the recursive Bayesian filter [3]. The initial position is assumed as the last GPS position fix. The main task of the algorithm would be to compensate for IMU data errors using map-matching. Our proposed algorithm is assumed to be a mobile-based technique, where map information is provided by network operators. However, the technique could be run as network-based if the IMU data is uploaded from the mobile terminal (MT) to the operator network. The objective of this paper is to investigate the feasibility of MT location using IMU raw data with cell-ID and geographical map information. We examine this concept by fusing simulat...

Citation Context

...he experimental area. We show how to maintain location information for mobile terminals in wireless networks using a novel combination of data sources. Our approach provides a reliable solution in street canyons and heavy tree canopies where GPS information is almost always inapplicable. The developed technique could also be applied to vehicle navigation, where dead-reckoning instruments are available and accurate. I. INTRODUCTION The first application of mobile location dates back to World War II, when it was critical to locate military personnel rapidly and precisely in emergency situations [1]. In the nineties, the GPS was made accessible for commercial applications. Furthermore, the EU is most likely to follow the US and Japan in requiring high positioning accuracy of mobile emergency calls from 2010 when the Galileo system will be fully operational [2]. However, the benefits of GPS could be limited where position information is still needed due to obscured view to satellites or degraded accuracy caused by multipath. A backup to GPS during signal outage with comparable accuracy could be achieved using fusion of inertial measurement unit (IMU) raw data with already existing cellID ...

Citation Context

...nsxx θ 111 sin −−− − ⋅+= tttt transyy θ Update Step for ni :1= do 22 )()( 1 itit i yyxx w −+− = −− endfor )( tt msortm = // Descending sort w.r.t weight ),(),( 11 yxyxs ttt == return( ts ) noisy t noisy ttt transxx 111 cos. −−− += θ , (8) noisy t noisy ttt transyy 111 sin. −−− += θ . (9) B. World Model Two kinds of databases (prior information) have been utilized in this work. The first one is a prediction of the radio profile in a test area of 9 km2 in Hannover, Germany. The predicted radio profile has been constructed using a 3D deterministic radio propagation prediction model, described in [5], with a resolution of 5 m. These data have been generated to provide predictions of the average received signal strength levels (RxLev), at reference locations, from the surrounding GSM antennas at 1800 MHz in our test area that contains 6 sectorized cells and four indoor antennas. This procedure is produced during the network planning stage, and is a useful source for MT positioning. After several preprocessing steps, as in [6] and [7], the radio profile data was subdivided into separate databases, in each are locations served by a certain cell antenna as illustrated in Figure 1. The second ...

Citation Context

...l states. However, normalization is not crucial for filter implementation. The term ),|( msop tt is the likelihood of the measurement or observation to of the serving cell-ID at time t given the current MT position and the world model m . It is also known as the observation model. The expression ),,|( 11 massp ttt −− is the probability that the MT is at ts given it executed the movement 1−ta when it was at 1−ts within the space defined by m . It is also called the motion model. Finally, )( 1−tsBel is the prior belief over the MT position. A complete derivation of expression (1) is provided in [4]. TABLE I shows how Equation (1) is usually computed in two steps called prediction and update, where )( tsBel − is the posterior belief just after executing action 1−ta and before incorporating the observation to . Note that MT actions and observations are assumed to occur in an alternative sequence. TABLE I. GENERIC RECURSIVE BAYESIAN FILTER B. Practical Implementation A single iteration of the positioning algorithm is given in TABLE II. The inputs are the initial position ),( 111 −−− = ttt yxs , the IMU data ),( 111 −−− = ttt transa θ , where 1−ttrans and 1−tθ are the translation (after twi...

Citation Context

... a test area of 9 km2 in Hannover, Germany. The predicted radio profile has been constructed using a 3D deterministic radio propagation prediction model, described in [5], with a resolution of 5 m. These data have been generated to provide predictions of the average received signal strength levels (RxLev), at reference locations, from the surrounding GSM antennas at 1800 MHz in our test area that contains 6 sectorized cells and four indoor antennas. This procedure is produced during the network planning stage, and is a useful source for MT positioning. After several preprocessing steps, as in [6] and [7], the radio profile data was subdivided into separate databases, in each are locations served by a certain cell antenna as illustrated in Figure 1. The second kind is a digital map of the area, generated from satellite images. Thus, different features, e.g. water, green, building, road, etc., could be easily discriminated. Because the goal of this work was to introduce a backup to GPS pedestrian positioning, we have extracted locations in which a walking person might exist and correlated their coordinates to the radio profile prediction data. The result is a collection of pedestrian ou...

Citation Context

...t always inapplicable. The developed technique could also be applied to vehicle navigation, where dead-reckoning instruments are available and accurate. I. INTRODUCTION The first application of mobile location dates back to World War II, when it was critical to locate military personnel rapidly and precisely in emergency situations [1]. In the nineties, the GPS was made accessible for commercial applications. Furthermore, the EU is most likely to follow the US and Japan in requiring high positioning accuracy of mobile emergency calls from 2010 when the Galileo system will be fully operational [2]. However, the benefits of GPS could be limited where position information is still needed due to obscured view to satellites or degraded accuracy caused by multipath. A backup to GPS during signal outage with comparable accuracy could be achieved using fusion of inertial measurement unit (IMU) raw data with already existing cellID based methods and map-matching. The radio profile of a given area can determine routes that are covered by each cell antenna. Therefore, the computational cost of map-matching algorithms would be reduced to a minimum. The proposed positioning algorithm is designed t...

Citation Context

...area of 9 km2 in Hannover, Germany. The predicted radio profile has been constructed using a 3D deterministic radio propagation prediction model, described in [5], with a resolution of 5 m. These data have been generated to provide predictions of the average received signal strength levels (RxLev), at reference locations, from the surrounding GSM antennas at 1800 MHz in our test area that contains 6 sectorized cells and four indoor antennas. This procedure is produced during the network planning stage, and is a useful source for MT positioning. After several preprocessing steps, as in [6] and [7], the radio profile data was subdivided into separate databases, in each are locations served by a certain cell antenna as illustrated in Figure 1. The second kind is a digital map of the area, generated from satellite images. Thus, different features, e.g. water, green, building, road, etc., could be easily discriminated. Because the goal of this work was to introduce a backup to GPS pedestrian positioning, we have extracted locations in which a walking person might exist and correlated their coordinates to the radio profile prediction data. The result is a collection of pedestrian outdoor lo...