System and method of traffic prediction for IoT nodes

1. A system of traffic prediction for Internet of Things (IoT) nodes, comprising at least one access node, a transmission network and a cloud platform; wherein

the at least one access node is configured to collect traffic data, cluster the traffic data according to path information in the traffic data to obtain access traffic data and network traffic data, input the access traffic data into an access traffic prediction model configured, output a prediction result of the access traffic at a next moment, and upload the network traffic data and the prediction result of the access traffic at the next moment to the cloud platform through the transmission network, and

the cloud platform is configured to input the network traffic data uploaded by the at least one access node into a network traffic prediction model configured, output a prediction result of the network traffic for each access node at a next moment, and obtain a prediction result of traffic for each node according to the prediction result of the network traffic at the next moment for each access node and the prediction result of the access traffic at the next moment received from each access node.

2. The system according to claim 1 , wherein the traffic data information comprises: date, time, duration, size, path information of the traffic and an ID of a node collecting information, wherein the path information comprises an IP address and port information of a source node and an IP address and port information of a destination node of the traffic;

after collecting the traffic data, the at least one access node stores the collected traffic data in a database in a format of a tuple; and

the at least one access node determines whether the traffic data is the access traffic data or the network traffic data based on the source node IP and the destination node IP in the path information of the traffic data.

3. The system according to claim 1 , wherein, the at least one access node compares the prediction result of the access traffic at the next moment with a traffic threshold for the access node corresponding to the current moment h t p , and uploads the prediction result of access traffic at the next moment and the network traffic data to the cloud platform when the prediction result of access traffic at the next moment is smaller than h t p .

4. The system according to claim 1 , wherein, when the prediction result of access traffic at the next moment is greater than or equal to h t p , the at least one access node backs up the network traffic data locally, and uploads the prediction result of access traffic at the next moment and a size of the network traffic data s t P to the cloud platform.

5. The system according to claim 1 , wherein, the access node further cleans incomplete data records, repeated data records and records roaming to the local in the traffic data information collected before clustering the traffic data.

6. The system according to claim 1 , wherein, the access node further converts the access traffic data into a data format of a training set of the access traffic prediction model, adds the access traffic data into the training set and performs a real-time training to update the access traffic prediction model.

7. The system according to claim 1 , wherein, the cloud platform further determines a traffic threshold for each access node corresponding to the next moment according to the prediction result of network traffic for each access node; and issues the traffic threshold for each access node corresponding to the next moment to the corresponding access node.

8. The system according to claim 7 , wherein, the cloud platform calculates a traffic threshold for a port n at an access node p corresponding to the next moment according to a prediction result of traffic r t+1 network,p,n to be forwarded at the next moment for the port n of the node p through the following equation:

h t+1 p,n =m 0 −Σ i r t+1 network,p,i , i∈I

wherein, m 0 is a physical maximum bearing traffic limit for the access node; I is a set of ports having path dependencies with the port n;

the cloud platform calculates the traffic threshold of the access node p corresponding to the next moment according to the traffic threshold for the port n of the access node p corresponding to the next moment through the following equation:

h t+1 p =Σ n h t+1 p,n , n∈N

wherein, N is a set of all ports of the access node p.

9. The system according to claim 1 , wherein the cloud platform further adds the network traffic data to a training set and performs a real-time training to update the network traffic prediction model.

10. The system according to claim 1 , wherein the cloud platform further determines a size of local data required to be uploaded to the cloud platform at the next moment for each access node, and issue the size of the local data required to be uploaded to the cloud platform at the next moment for each access node to the corresponding access node;

the cloud platform determines the prediction result of traffic for each node according to the prediction result of the network traffic at the next moment for each access node, the prediction result of the access traffic at the next moment received from each access node and the size of the local data required to be uploaded to the cloud platform at the next moment for each access node.

11. A method of traffic prediction for IoT, comprising:

collecting, by each access node, traffic data;

clustering, by each access node, the traffic data into access traffic data and network traffic data according to path information in the traffic data;

inputting, by each access node, the access traffic data into an access traffic prediction model configured, to obtain a prediction result of the access traffic at a next moment output by the traffic prediction model;

uploading, by each access node, the prediction result of access traffic at the next moment and the network traffic to a cloud platform;

receiving, by the cloud platform, the network traffic data and the prediction result of access traffic at the next moment reported by each access node;

inputting, by the cloud platform, the network traffic data into a network traffic prediction model configured, to obtain a prediction result of the network traffic for each access node at the next moment output by the network traffic prediction model; and

determining and outputting, by the cloud platform, a prediction result of traffic at the next moment for each access node according to the prediction result of access traffic at the next moment for each access node and the prediction result of network traffic at the next moment.

12. The method according to claim 11 , wherein the traffic data information comprises:

date, time, duration, size, path information of the traffic and an ID of a node collecting information, wherein the path information comprises an IP address and port information of a source node and an IP address and port information of a destination node of the traffic;

the method further comprises: storing, by the at least one access node, the collected traffic data in a database in a format of a tuple, after collecting the traffic data; and

clustering the traffic data into access traffic data and network traffic data according to path information in the traffic data comprises: determining whether the traffic data is the access traffic data or the network traffic data based on the source node IP and the destination node IP in the path information of the traffic data.

13. The method according to claim 11 , further comprising:

comparing, by the at least one access node, the prediction result of the access traffic at the next moment with a traffic threshold for the access node corresponding to the current moment h t p ; and

uploading the prediction result of access traffic at the next moment and the network traffic data to the cloud platform when the prediction result of access traffic at the next moment is smaller than h t p .

14. The method according to claim 11 , further comprising:

backing up, by the at least one access node, the network traffic data locally when the prediction result of access traffic at the next moment is greater than or equal to h t p ; and

uploading the prediction result of access traffic at the next moment and a size of the network traffic data s t P to the cloud platform.

15. The method according to claim 11 , further comprising:

cleaning, by the access node, incomplete data records, repeated data records and records roaming to the local in the traffic data information collected before clustering the traffic data.

16. The method according to claim 11 , further comprising:

converting, by the access node, the access traffic data into a data format of a training set of the access traffic prediction model;

adding, by the access node, the access traffic data into the training set; and

performing, by the access node, a real-time training to update the access traffic prediction model.

17. The method according to claim 11 , further comprising:

determining, by the cloud platform, a traffic threshold for each access node corresponding to the next moment according to the prediction result of network traffic for each access node; and

issuing, by the cloud platform, the traffic threshold for each access node corresponding to the next moment to the corresponding access node.

18. The method according to claim 17 , further comprising:

calculating, by the cloud platform, a traffic threshold for a port n at an access node p corresponding to the next moment according to a prediction result of traffic r t+1 network,p,n to be forwarded at the next moment for the port n of the node p through the following equation:

h t+1 p,n =m 0 −Σ i r t+1 network,p,i , i∈I

wherein, m 0 is a physical maximum bearing traffic limit for the access node; I is a set of ports having path dependencies with the port n;

calculating, by the cloud platform, the traffic threshold of the access node p corresponding to the next moment according to the traffic threshold for the port n of the access node p corresponding to the next moment through the following equation:

h t+1 p =Σ n h t+1 p,n , n∈N

wherein, N is a set of all ports of the access node p.

19. The method according to claim 11 , further comprising:

adding, by the cloud platform, the network traffic data to a training set; and

performing, by the cloud platform, a real-time training to update the network traffic prediction model.

20. The method according to claim 11 , further comprising:

determining, by the cloud platform, a size of local data required to be uploaded to the cloud platform at the next moment for each access node;

issuing, by the cloud platform, the size of the local data required to be uploaded to the cloud platform at the next moment for each access node to the corresponding access node; wherein,

determining a prediction result of traffic at the next moment for each access node comprises: determining the prediction result of traffic for each node according to the prediction result of the network traffic at the next moment for each access node, the prediction result of the access traffic at the next moment received from each access node and the size of the local data required to be uploaded to the cloud platform at the next moment for each access node.