Research On Angle-of-arrival Based Indoor Localization

Indoor localization techniques have attracted much attention in recent years due to the widespread indoor location-based service(ILBS)in social and commercial fields,such as indoor navigation and security surveillance.The poor penetration of the GPS signal promotes the exploration of some new technologies.The localization techniques based on Wi-Fi have been shown to be competitive due to the ubiquitousness of Wi-Fi access points(APs)and the high accuracy.Particularly,the angle-of-arrival(Ao A)based Wi-Fi system is easy to implement and can achieve the localization accuracy of tens of centimeters.Utilizing the triangulation,the Ao A system collects the Ao A estimates from the APs to calculate the target position.However,the current localization system is vulnerable to the obstacles and array deployment.First,the existence of obstacle will block the line-of-sight between the AP and target.An Ao A estimate from the non-line-of-sight AP will degrade the localization performance.Second,It is analyzed that the Ao A estimation accuracy is related to the orientation of the array that is used to estimate the Ao A.The perpendicular incident wave contributes to the highest Ao A estimation accuracy.Accordingly,how to utilize the array orientation to build a high-accuracy for an obstacle-aware scenario is quite important.In this paper,we first analyze the impact of the obstacles and array orientation on the localization performance.On this basis,we carry out the system design,targeting at deploying the minimum number of APs to establish a high-accuracy localization system.The main contributions are listed as follows.1.We carry out the high-accuracy localization system using the minimum number of APs,where the array orientation,the obstacles and path loss are considered.First,we utilize the Cramer-Rao lower bound(CRLB)to quantify the performance limit for the localization system,which characterizes the effect of target-AP geometry,array orientation,obstacles and path loss on the localization error.Exploiting the additive property of the Fisher information matrix(FIM,the inverse of CRLB),the system design problem is formulated as an AP selection problem.By gridding the AP candidates,we aim to select the minimum number of APs under the constraints of keeping the localization error below a given limit.Aided by the convex relaxation,we design a heuristic algorithm to alternatively chooses the most informative AP until the performance for each target is satisfied.2.We design an array optimization method.The array optimization scheme can flexibly reshape the array orientation through array rotation to improve the Ao A estimation accuracy.Appropriate array orientation efficiently reduces the uncertainty in Ao A estimation,thereby improving the localization performance.The array deployment strategy is related to the AP number and AP positions.For the case with abundant APs,the array deployment method optimizes the array orientation for each AP,which makes sure that the Ao A estimate obtained at each AP is accurate.For the two-AP case,we make the array irregularly deployed to combat the localization error caused by the linear array symmetry.Irregular deployment gives more Ao As and eliminates the linear array symmetry,which decreases the localization error.3.We design a localization algorithm which is robust to the array deployment.Although the phased array is designed in a linear and equidistant pattern,these conditions cannot be strictly satisfied in actual deployments.This kind of systematic error is small but nonnegligible.Little shift of the antenna arrangement will result in the biased Ao A estimation error.Accordingly,we devise an array calibration method to cope with irregular arrays.The calibration method incorporates a fitting algorithm to find the little shift in antenna spacing or array orientation.With these deviations,we develop an improved Ao A estimation algorithm.The improved algorithm is able to accommodate the deployment errors and accurately estimate the Ao A even with irregular arrays,making Rc Loc robust to the antenna deployment error.Furthermore,we propose a weighted Ao A-based localization method,in which the accuracy of each estimated Ao A is considered.