Kailiang Zhang

Bio: Kailiang Zhang is an academic researcher from Xuzhou Institute of Technology. The author has contributed to research in topics: Quality of experience & Cloud computing. The author has an hindex of 2, co-authored 9 publications receiving 10 citations.

A Hyperparameters automatic optimization method of time graph convolution network model for traffic prediction

Abstract: Smart transportation is an essential component of the smart city. Traffic prediction is an important issue in smart transportation. The convolutional neural networks (GCN) are an effective approach for traffic prediction. However, the GCN meets some challenges, such as stability of prediction precision and computation cost, in traffic prediction. The hyperparameters significantly affect the performance of GCN. We conduct a regression analysis between hyperparameters and GCN performance. Our simulation results show that there is the obvious optimal point of hyperparameters. Some empirical suggestion is given to adjust the hyperparameters based on the simulation results.

A Test System for Vehicular Voice Cloud Service

Abstract: Some test systems for voice cloud services have been developed in recent years. However, the automobile manufacturers, communications equipment merchants and network operator still lacks methods and tools to evaluate the vehicular voice cloud services from the perspective of the end user experience. Considering the user behavior and user experience, a light weight vehicular voice cloud evaluation system is designed in this paper. The system is able to send voice information to voice cloud server according to user habit, and record the user experience indicators, such as accurate, voice quality, service delay, server computation capacity, and so on. The study shows that the vehicle voice cloud evaluation system can avoid complex communication and language processing, evaluate the performance of the service from view of end user.

Bursty data service latency analysis under fractional calculus fluid model of Multi-hop Wireless Networks

Abstract: A fractional calculus fluid model can be used to better explain the bursty data service traffic, which is long-range dependence and has a fractal like the feature of network data flow. The heavy-tailed delay distributions, the hyperbolic decay of the packet delay auto-covariance function and fractional differential equations are shown to be formally related. Effective capacity is a useful model to describe wireless networks with QoS constraints. This paper builds a fluid model to describe the traffic of multi-hop wireless networks under QoS constraints. The proposed method could analyze the relationship between latency and a complicated traffic model, which is more similar to the real scenario.

Research on Image Data Clustering Algorithm Based on Low Rank Subspace Clustering

Abstract: At present, the scale and types of data collected by people have shown explosive growth. It is very difficult to obtain specific and effective classification labels for high-dimensional data. By using subspace clustering method with low rank representation, the linear representation matrix of the data with the lowest rank is found, and the global structure of the original data is preserved to achieve the purpose of optimizing clustering. By comparing K-means clustering, LRR clustering and the improved LRR clustering method of self-adapting graph regularization low rank representation, the experiment proves that the latter has better effect in clustering image data collected from different angles.

Electric pile power supply control system is filled to intelligence environmental protection formula

Abstract: The utility model relates to an electric pile power supply control system is filled to intelligence environmental protection formula belongs to the electric pile control system technical field of filling, including microprocessor, display screen, relay drive switch, alarm module, communication module, solar energy supply circuit, battery supply circuit, commercial power supply circuit, step -downdetection circuitry and multistage adjustable voltage output, rational utilization solar energy can be stored the electric energy in the battery in sunny, utilizes the battery to supply power for equipment when sunshine is deficient, and does not also have under the condition of electricity at the battery, can select the commercial power to be its power supply, and is energy -concerving and environment -protective, has very high economic benefits, centralized management in time understands the running state who fills electric pile, has good man -machine interface, and reaches the server on canwill filling each item operating parameter of electric pile through the communication module who carries, makes things convenient for that managers in time understands, centralized management.

Massive MIMO for Maximal Spectral Efficiency: How Many Users and Pilots Should Be Allocated?

Abstract: Massive MIMO is a promising technique for increasing the spectral efficiency (SE) of cellular networks, by deploying antenna arrays with hundreds or thousands of active elements at the base stations and performing coherent transceiver processing. A common rule-of-thumb is that these systems should have an order of magnitude more antennas $M$ than scheduled users $K$ because the users’ channels are likely to be near-orthogonal when $M/K > 10$ . However, it has not been proved that this rule-of-thumb actually maximizes the SE. In this paper, we analyze how the optimal number of scheduled users $K^\star$ depends on $M$ and other system parameters. To this end, new SE expressions are derived to enable efficient system-level analysis with power control, arbitrary pilot reuse, and random user locations. The value of $K^\star$ in the large- $M$ regime is derived in closed form, while simulations are used to show what happens at finite $M$ , in different interference scenarios, with different pilot reuse factors, and for different processing schemes. Up to half the coherence block should be dedicated to pilots and the optimal $M/K$ is less than 10 in many cases of practical relevance. Interestingly, $K^\star$ depends strongly on the processing scheme and hence it is unfair to compare different schemes using the same $K$ .

A Hyperparameters automatic optimization method of time graph convolution network model for traffic prediction

Abstract: Smart transportation is an essential component of the smart city. Traffic prediction is an important issue in smart transportation. The convolutional neural networks (GCN) are an effective approach for traffic prediction. However, the GCN meets some challenges, such as stability of prediction precision and computation cost, in traffic prediction. The hyperparameters significantly affect the performance of GCN. We conduct a regression analysis between hyperparameters and GCN performance. Our simulation results show that there is the obvious optimal point of hyperparameters. Some empirical suggestion is given to adjust the hyperparameters based on the simulation results.

A Survey of Image Clustering: Taxonomy and Recent Methods

Abstract: Image clustering is a fundamental problem in computer vision domains. In this survey, we provide a comprehensive overview of image clustering. Specifically, we first discuss the applications of image clustering across various domains. Then, we summarize the common algorithms and propose a classification of image clustering. The existing methods are classified from four aspects: autoencoder based methods, subspace clustering, graph convolution network (GCN) based methods and some other clustering methods. We introduce the main research contents and existing problems of various image clustering methods. We also introduce some recent methods and summarize the experimental results. Based on our taxonomy and analysis, creating and verifying new methods is more straightforward. Finally, we propose the future opportunities in this fast developing field.