40results about How to "Reduce storage redundancy" patented technology

The present invention proposes a user priority-oriented base station cooperative caching method in a dense scene, which mainly solves the low hit rate of user request content in the base station cache in the prior art and the high priority caused by user requests being served without differentiation The problem of users being treated unfairly. The implementation plan is: first establish a deep reinforcement learning model for base station cooperative caching according to the request information of all users and the information in the base station buffer; then conduct offline training and learning on the model to obtain a learned deep reinforcement learning model; finally use The learned deep reinforcement learning model performs online base station cooperative caching decision at the macro base station. The invention not only increases the hit rate of the user's request content in the base station buffer, but also provides differentiated services to users with different priorities, which can not only meet the Internet access requirements of different users, but also bring more benefits to operators. Great benefits can be used for user access to resources in wireless networks.

The present invention discloses a method for storing dimension data, comprising: acquiring dimension data newly generated by terminal operation; determining whether the newly generated dimension data satisfies storage conditions according to a pre-configured dimension strategy, and the dimension strategy is judging Whether the dimension data satisfies the storage condition; when the storage condition is satisfied, the newly generated dimension data is added to the dimension data sequence for storage. The method for storing dimensional data provided by the embodiment of the present invention only stores data that meets the storage conditions, thereby reducing the storage capacity of time series data and reducing storage redundancy.

The invention discloses a document processing method taking a lexical chunk as a basic unit. The method comprises the following steps: establishing a document item and a catalog outline of the document item, wherein the catalog outline is used for recording catalog information and catalog structure; adding a plurality of lexical chunks under the catalog outline according to the catalog structure of the catalog outline and editing the lexical chunks, wherein the lexical chunks all have attribute marks and are stored in a same lexical chunk bank and the attribute marks include a value mark, a content mark, a semantic mark, a position mark and a citing state mark; storing the document item which takes the lexical chunk as the basic unit. The document processing method disclosed by the embodiment of the invention can be adopted for reducing storage redundancy and is convenient for online editing of the same document by a plurality of persons.

The invention discloses a distributed storage transformer station inspection robot remote centralized monitoring system and method. The system includes a plurality of in-station servers and a plurality of remote servers, the in-station servers are deployed in the corresponding substations and receive the data of an in-station network video server and a substation inspection robot respectively, the substation inspection robot transmits the video data to the in-station network video server, the in-station network video server receives all the video data in the substation, the in-station servers upload the data to the corresponding remote servers, and the in-station network video server uploads the data to the plurality of remote servers, and the remote servers receive the data uploaded by the in-station servers and the network video server in the substation corresponding to the remote servers, and upload the data to the corresponding remote centralized control clients. The system and method overcome the problem that the unified storage mode is large in data bulk, the retrieval speed is slow and the resources are consumed.

The present invention relates to the technical field of data processing, in particular to an HBase secondary index query and storage system and query method. The HBase secondary index query and storage system comprises a client, a query processing module, a query execution engine module and an HBase storage module. According to the HBase secondary index query and storage system and query method, filter columns originally stored in each index table are combined into a main filter column table to enable filter column data requiring redundancy to be greatly reduced, thereby reducing storage redundancy of data; and meanwhile, the queries to all index tables are finally integrated into the query to the main filter column table through a mixer filter, the main filter column table is stored in a rapid storage medium at the same time, and the query efficiency is relatively high; moreover, the query efficiency is greatly improved by using the rapid random access performance of an SSD hard disc.

The invention provides a data management method and device. The data management method can perform trajectory storage and querying by using a trajectory pattern extracted from lots of historical trajectories. The data management method includes: matching original trajectory points to a road network through a map matching method, and acquiring a trajectory set after map matching; determining a minimum support and a minimum road segment number corresponding to different road types according to training data; establishing a trajectory tree according to the trajectory set, and establishing a trajectory sub-tree with each node of the trajectory tree as a staring point, wherein the trajectory tree and the trajectory sub-trees include at least one node, and each node is an intersection of the road network; determining a trajectory pattern meeting the minimum support and the minimum road segment number, passing through each node according to the trajectory sub-trees; and performing redundancy elimination treatment on the trajectory pattern of each node, and generating a new trajectory pattern and storing the new trajectory pattern. The data management method and device can reduce trajectory storage redundancy, and can accelerate the trajectory querying speed.

The invention provides a mass file storage method and device for a distributed type system. The method comprises the steps of encoding 2n data files to be stored to generate m global verifying block files; dividing the 2n data files into k groups, encoding each group of data files to generate corresponding local verifying block files; storing the 2n data files, the m global verifying block files and the k local verifying block files in 2n+m+k severs in the distributed type system according to a mode that only one file is stored in one server. According to the scheme, encoding treatment is conducted on original data, the global verifying block files and the local verifying block files are calculated, then, original data files and the verifying block files simply need to be stored in the distributed type system instead of storing multiple copies of the original data files, and thus the usage rate of storage equipment is obviously increased on the premise of guaranteeing that the storage reliability is not lowered.

The invention discloses a block chain-based medical data management system, which is used to solve the problem that the existing medical data management system does not screen the medical data, which leads to the phenomenon that the data is found, but the data quality is found to be very poor and the medical data is not checked. Carry out reasonable storage and reduce the problem of block chain storage node storage redundancy; including medical data collection module, data encryption module, node distribution module, server, node collection module and block chain node module; the present invention uses node storage value to medical treatment The data is screened to select high-quality medical data for storage; the matching value of the blockchain storage node is obtained through information collection and analysis of the blockchain storage node, and the blockchain storage node is reasonably selected through the matching value Storage; through the management value, delete and compress the data assigned by the node, thereby reducing the storage redundancy of the blockchain storage node.