University of Electronic Science and Technology of China

Jing Ren

Sheng Wang

Shizhong Xu

Hongfang Yu

Du Xu

Gang Sun

Qing Li

Yuhui Wang

Tongyu Song

Xiaotong Guo

Xiong Wang

Jiangong Zheng

Xuebin Tan

Jian Sun

A method for energy efficient routing in wireless sensor network based on multi-agent deep reinforcement learning, predefines a to-be-deployed wireless sensor network and creates a cooperative routing decision system including A decision networks and one sink module, A decision networks deployed on the agents a, i=1, 2, . . . , A, of the sensor nodes, the sink module deployed on the sink node n. The decision network obtains a probability vector according to its local observation and position vectors. The sink module calculates a routing for each sensor node according the probability vectors of A decision networks and sends the routings to corresponding sensor nodes. A multi-agent deep reinforcement learning algorithm is adopted to train the decision networks of A agents a, i=1, 2, . . . , A of the cooperative routing decision system, deploys the to-be-deployed wireless sensor network into an environment and updates the routing policy of the deployed wireless sensor network at each update cycle of routing.

Method for energy efficient routing in wireless sensor network based on multi-agent deep reinforcement learning

A dynamic network resource allocation method based on network slicing is provided. A historical resource demand dataset of an accessed network slice is inputted into a first neural network for training. Based on a trained first neural network and the historical resource demand of the accessed network slice, a resource demand prediction information corresponding to the accessed network slice in a first prediction time period is determined. Resources are pre-allocated to the accessed network slice based on the resource demand prediction information, and resources are allocated to the accessed network slice when the first prediction time period arrives. In this way, the service provider can reasonably allocate network resources for network slices without violating the SLA, thus avoiding the waste of network resources.

Dynamic network resource allocation method based on network slicing

The present invention relates to a method for detecting Network Anomaly in network architectures based on locator/identifier split, the detection flow is as follows: initialization processing, and in ITR: processing data packets, sending a Map-Request, determining whether to send an additional Map-Request, sending the data packet, processing the Map_Reply, processing EID-to-RLOC Cache entry expired; in ETR: processing data packet, processing Map-Request, determining whether the traffic of the ITR currently sending the Map-Request is abnormal, replying to the ITR of which the query traffic is abnormal, replying to ITR of which the query traffic is abnormal. With respect to the characteristic that the network architecture based on locator/identifier split needs to query the relationship between the locator and the identifier for packet delivery, the present invention detects Network Anomaly based on query traffic instead of network data packet traffic. Thus the present invention has the advantages of effectively reducing the investment on detection device, The overhead of exchanging monitoring information and the detection system maintenance cost; facilitating cross domain coordination; and efficiently handling the failures occurring during network operation in time; effectively improve the reliability of the network, being suitable for a large-scale network.

Method for network anomaly detection in a network architecture based on locator/identifier split

Growing community of inventors

Chengdu, China

Jing Ren