We analyze the decade-long (2007-2016) record of aerosol measurements from four distinctive sensors, namely CALIOP, MODIS, MISR, and IASI, to quantify the transAtlantic dust transport and deposition. These satellite sensors use different techniques to characterize particle size and shape properties; and we have developed sensor-specific methods (broadly categorized into size-based and shape-based method) to derive dust optical depth (DOD). The size-based DOD from MODIS and IASI generally agrees better with AERONET -derived DOD than the shape-based DOD from CALIOP and MISR does. Overall, the shape-based DOD is smaller than the size-based DOD by about 25%, which is consistent with distinctive ways of accounting for coarse-spherical particles of dust-pollution internal mixture. While such dust-pollution mixtures are counted as dust in the size-based DOD, they are excluded in the shape-based DOD. DOD is not a good proxy for dust deposition. Instead, the dust deposition depends strongly on the gradient of DOD on a monthly basis and can be derived by calculating the meridional and zonal dust mass flux based on the three-dimensional distributions of dust. Among the remote sensing measurements, difference in dust deposition is smaller than that of DOD, suggesting that different satellites characterize the DOD gradient more consistently than DOD itself. Satellite measurements of dust deposition and DOD also provide an accurate estimate of the dust loss frequency (LF) that measures how efficient the dust is removed from the atmosphere. We found that these remote sensing measurements yield similar LF values of 0.078 - 0.102 d-1, which however is factors of 2-5 smaller than model simulations. This analysis provides valuable insights into potential deficiencies in models' emission and transport/removal processes and hence helps guide model improvement.

Detection and tracking of moving objects is one of the most essential functions of autonomous cars. In order to estimate the dynamic information of a moving object accurately, laser scanners are widely used for their highly accurate distance data. However, these data only represent the surface of an object facing the sensor and changes the appearance of an object over time. This change produces unexpected tracking errors of estimated dynamic states. In this paper, in order to minimize the tracking error caused by appearance changes, a tracking algorithm based on L-shaped model switching is proposed. The suggested algorithm is validated in real traffic experiments where position, velocity, and heading angle error were measured by using precise GPS. The L-shape tracking algorithm successfully mitigated the effect of appearance changes and improved estimation performance.

In many real applications such as bioinformatics and biological network analysis, it has always been an important, yet challenging, topic to accurately infer/reconstruct gene regulatory networks (GRNs) from microarray data, and efficiently identify those matching GRNs with similar interaction structures for potential disease analysis and treatment tasks. Motivated by this, in this paper, we formalize the problem of ad-hoc inference and matching over gene regulatory networks (IM-GRN), which deciphers ad-hoc GRN graph structures online from gene feature databases (without full GRN materializations), and retrieves the inferred GRNs that are subgraph-isomorphic to a query GRN graph with high confidences. Specifically, we propose a novel probabilistic score to measure the possible interaction between any two genes (inferred from gene feature vectors), and thus model GRNs by probabilistic graphs, containing edge existence probabilities. In order to efficiently process IM-GRN queries, we propose effective reduction, pruning, and embedding strategies to significantly reduce the search space of GRN inference and matching, without materializing all GRNs. We also present an effective indexing mechanism and an efficient IM-GRN query processing algorithm by the index traversal. Finally, extensive experiments have been conducted to verify the efficiency and effectiveness of our proposed IM-GRN query answering approaches over real/synthetic GRN data sets.

This paper proposes a multiple exposure images based traffic light recognition method. For traffic light recognition, color segmentation is widely used to detect traffic light signals; however, the color in an image is easily affected by various illuminations and leads to incorrect recognition results. In order to overcome the problem, we propose the multiple exposure technique which enhances the robustness of the color segmentation and recognition accuracy by integrating both low and normal exposure images. The technique solves the color saturation problem and reduces false positives since the low exposure image is exposed for a short time. Based on candidate regions selected from the low exposure image, the status of six three and four bulb traffic lights in a normal image are classified utilizing a support vector machine with a histogram of oriented gradients. Our algorithm was finally evaluated in various urban scenarios and the results show that the proposed method works robustly for outdoor environments.

In this paper, a new efficient and accurate analytical eye-diagram determination technique for interconnect lines is presented. The simplest input test signal model for the intersymbol interference analysis of high-speed data links is mathematically formulated. Since input test patterns for eye boundaries are determined analytically, it is considered very convenient and efficient. The proposed technique shows excellent agreement with the SPICE-based simulation in both eye height and jitter, i.e., within 5% error for nondiscontinuous data paths and 10% error for discontinuous data paths. The method is much more computation-time-efficient than the pseudorandom bit sequence-based SPICE simulation in the order of magnitude.

An efficient signal integrity verification method of coupled interconnect lines is presented. Two coupled lines are decoupled into isolated eigen modes. Then, bit blocks for coupled lines which are composed of the finite size of bits are represented with the fundamental modes. In addition, the crosstalk effects within the bit block are taken into account. Thereby, the crucial input bit patterns for the worst case eye-diagram determination are mathematically modeled, followed by analytical eye-diagram determination. It is shown that not only the proposed technique has excellent agreement with SPICE W-model-based simulation but also is it very computation-time-efficient, compared with SPICE simulation.