An Improved Convolutional Neural Network Based Indoor Localization by Using Jenks Natural Breaks Algorithm

With the rapid growth of the demand for indoor location-based services(LBS), Wi-Fi received signal strength(RSS) fingerprints database has attracted significant attention because it is easy to obtain. The fingerprints algorithm based on convolution neural network(CNN) is often used to improve indoor localization accuracy. However, the number of reference points used for position estimation has significant effects on the positioning accuracy. Meanwhile, it is always selected arbitraily without any guiding standards. As a result, a novel location estimation method based on Jenks natural breaks algorithm(JNBA), which can adaptively choose more reasonable reference points, is proposed in this paper. The output of CNN is processed by JNBA, which can select the number of reference points according to different environments. Then, the location is estimated by weighted K-nearest neighbors(WKNN). Experimental results show that the proposed method has higher positioning accuracy without sacrificing more time cost than the existing indoor localization methods based on CNN.

Population spatialization with pixel-level attribute grading by considering scale mismatch issue in regression modeling

Population spatialization is widely used for spatially downscaling census population data to finer-scale.The core idea of modern population spatialization is to establish the association between ancillary data and population at the administrative-unit-level(AUlevel)and transfer it to generate the gridded population.However,the statistical characteristic of attributes at the pixel-level differs from that at the AU-level,thus leading to prediction bias via the cross-scale modeling(i.e.scale mismatch problem).In addition,integrating multi-source data simply as covariates may underutilize spatial semantics,and lead to incorrect population disaggregation;while neglecting the spatial autocorrelation of population generates excessively heterogeneous population distribution that contradicts to real-world situation.To address the scale mismatch in downscaling,this paper proposes a Cross-Scale Feature Construction(CSFC)method.More specifically,by grading pixel-level attributes,we construct the feature vector of pixel grade proportions to narrow the scale differences in feature representation between AU-level and pixel-level.Meanwhile,fine-grained building patch and mobile positioning data are utilized to adjust the population weighting layer generated from POI-density-based regression modeling.Spatial filtering is furtherly adopted to model the spatial autocorrelation effect of population and reduce the heterogeneity in population caused by pixel-level attribute discretization.Through the comparison with traditional feature construction method and the ablation experiments,the results demonstrate significant accuracy improvements in population spatialization and verify the effectiveness of weight correction steps.Furthermore,accuracy comparisons with WorldPop and GPW datasets quantitatively illustrate the advantages of the proposed method in fine-scale population spatialization.