Search

VLC-WIFI heterogeneous networks have become a promising solution, where VLC and WIFI are collaborating on providing services for indoor users. The dense deployment of visible light communication multiple access points (APs) can expand the system capacity, but it leads to complicated inter-cell interference (ICI). In order to arrange users to access the appropriate AP and implement interference management, a novel scheme called user-centric access based on regular spectrum allocation (UCARSA) is proposed in the paper. In UCARSA, the sliced visible spectrum is assigned to VLC APs regularly, so that the user associated with more than one AP is impacted less by ICI. Furthermore, the virtual community is constructed for each user to balance the user rate and improve the average spectral efficiency. The above performance improvement can be proved by simulation results in this paper, and it shows that the performance of UCARSA is better than two kinds of comparison algorithms.

Multi-agent reinforcement learning (MARL) has been gaining extensive attention from academia and industries in the past few decades. One of the fundamental problems in MARL is how to evaluate different approaches comprehensively. Most existing MARL methods are evaluated in either video games or simplistic simulated scenarios. It remains unknown how these methods perform in real-world scenarios, especially multi-robot systems. This paper introduces a scalable emulation platform for multi-robot reinforcement learning (MRRL) called SMART to meet this need. Precisely, SMART consists of two components: 1) a simulation environment that provides a variety of complex interaction scenarios for training and 2) a real-world multi-robot system for realistic performance evaluation. Besides, SMART offers agent-environment APIs that are plug-and-play for algorithm implementation. To illustrate the practicality of our platform, we conduct a case study on the cooperative driving lane change scenario. Building off the case study, we summarize several unique challenges of MRRL, which are rarely considered previously. Finally, we open-source the simulation environments, associated benchmark tasks, and state-of-the-art baselines to encourage and empower MRRL research., Comment: 13 pages

Object detection is an essential technique for autonomous driving. The performance of an object detector significantly degrades if the weather of the training images is different from that of test images. Domain adaptation can be used to address the domain shift problem so as to improve the robustness of an object detector. However, most existing domain adaptation methods either handle single target domain or require domain labels. We propose a novel unsupervised domain classification method which can be used to generalize single-target domain adaptation methods to multi-target domains, and design a weather-invariant object detector training framework based on it. We conduct the experiments on Cityscapes dataset and its synthetic variants, i.e. foggy, rainy, and night. The experimental results show that the object detector trained by our proposed method realizes robust object detection under different weather conditions.

Taihu Lake area is one of the most developed areas in agricultural production. Application of fertilizers and pesticides in large quantities greatly aggravate environmental pollution of this area, and water pollution has worsened to an unbearable condition. Two sampling farms (respectively 1 hm2) under rape-rice rotation and wheat-rice rotation were selected as studied ecosystem and a 5-yr-old Poplar forest and 8-yr-old Metasequoia forest were chosen in the selected areas. By collecting samples of Nitrogen, Phosphorus in water, crops and underground of forest, the transfer and loss of N and P (main water pollutants) in faming ecosystem were studied, and the effects of forest belts on non-point source pollution of agricultural lands was analyzed. The results indicated that the transfer and loss of N and P vary with means of rotation, types of crops and the amount of fertilizer application. Buffering forest belts betweens farmlands and ditches can effectively stop and purify such elements as N and P in soil runoffs, thus controlling non-point source pollution of agricultural lands. When the width ratio of farmland to forest belt is 100 to 40, 50.05% losing N, 29.37% losing P can be absorbed by forest under rape-rice rotation and 30.98% N, 86.73% P can be absorbed by forest under wheat-rice rotation. When the width ratio of farmland to forest belt is 150 to 40, 33.37% losing N, 19.58% losing P can be absorbed by the forest under rape-rice rotation, and under wheat-rice rotation 20.65% lost N and 57.82% lost P can be absorbed. There is only some purification effect when the width ration of farmland to forest belt is 200 to 40. Based on model of buffering forest belts, the width ratio of farmland to forest is determined between 100 to 40 and 150 to 40, because it not only can purify water, but also occupy less farmland. It is suggested that Poplars, with the characteristics of fast-growing and high value, are suitable to be planted as shelter-forest in Taihu Lake Watershed.

This paper presents a site selection algorithm for water resources points based on GIS technology. By analyzing the relationship between supply and demand constraints, it plans points-site and solves the water resources allocation problem. The main idea of this algorithm is as follows. Consider the situation of current water resources supply, define new points-sites where settlements are insufficient of water and calculate their service capabilities on the spatial constrains. By iteratively picking the point-sites with maximum service capabilities and updating the situation of supply and demand, a solution can be found with this greedy strategy. Although suboptimal, the algorithm has the merit of simplicity and concision. Water resources distribution system using this algorithm is implemented on the base of real geographic data from Qinghai province. Results of the application and discussion are provided in this paper.