Methods, systems, and computer program products for optimizing a predictive model for mobile network communications based on historical context information
Methods and systems are described for optimizing a predictive model for mobile network communications based on historical context information. In one aspect, historical context information is collected including at least one of communication environment, communication parameter estimates, mobile device statistics, mobile device transmit settings, base station receiver settings, past network statistics and settings, and adjacent network node information statistics and settings, the historical context information including data from communications of at least one mobile device. A predictive model for network communications is determined based on the historical context information. A communication context for a first mobile device different than the at least one mobile device is determined. The first device is scheduled and/or network parameters are set based on the determined predictive model and communication context.
1. A method for optimizing mobile network communications using network learning, the method comprising:
collecting historical context data representing performance of a plurality of mobile devices operating in a plurality of cells of a mobile network,
wherein the historical context data is collected in the plurality of cells of the mobile network, and
wherein the historical context data includes at least one of communication environment, communication parameter estimates, mobile device statistics, mobile device transmit settings, base station receiver settings, past network statistics and settings, and adjacent network node information statistics and settings;
building a predictive multi-cell model to optimize scheduling of network traffic in one or more cells of the mobile network based on the historical context data for the plurality of cells;
determining a first predictive cell-level model for a first cell based at least in part on the predictive multi-cell model;
determining a second predictive cell-level model for a second cell based at least in part on the predictive multi-cell model;
scheduling network traffic for a mobile device in the first cell using the first predictive cell-level model for the first cell;
scheduling network traffic for a mobile device in the second cell using the second predictive cell-level model for the second cell;
adjusting the first predictive cell-level model to further optimize scheduling of network traffic for the first cell based at least in part on additional information from the first cell;
adjusting the second predictive cell-level model to further optimize scheduling of network traffic for the second cell based at least in part on additional information from the second cell.
2. The method of claim 1 , wherein the predictive cell-level model for the first cell is based on the historical context data collected in the first cell.
3. The method of claim 2 , wherein the predictive cell-level model for the first cell is further based on the historical context data collected from cells other than the first cell.
4. The method of claim 1 , wherein the additional information from the first cell is collected from a mobile device operating in the first cell.
5. The method of claim 1 , wherein the additional information from the first cell is collected from a base station for the first cell.
6. The method of claim 1 , wherein the additional information from the first cell includes signal-to-interference-plus-noise ratio (SINR) for the first cell.
7. The method of claim 1 , wherein the additional information from the first cell is received in real time.
8. The method of claim 1 , wherein adjusting the first predictive cell-level model to further optimize scheduling of network traffic for the first cell based at least in part on the additional information from the first cell occurs in real time.
9. The method of claim 1 , wherein the plurality of cells of the mobile network are divided into multiple sub-networks.
10. The method of claim 9 , wherein a first sub-network of the multiple sub-networks is operated independently of a second sub-network of the multiple sub-networks.
11. The method of claim 9 , wherein network settings for a first sub-network of the multiple sub-networks differ from network settings for a second sub-network of the multiple sub-networks.
12. The method of claim 1 , wherein different predictive cell-level models for different cells optimize latency of the network traffic, and wherein the latency is optimized differently by the different predictive cell-level models for different cells.
13. The method of claim 1 , wherein different predictive cell-level models for different cells optimize bandwidth of the network traffic, and wherein the bandwidth is optimized differently by the different predictive cell-level models for different cells.
14. The method of claim 1 , wherein different predictive cell-level models for different cells optimize reliability of the network traffic, and wherein the reliability is optimized differently by the different predictive cell-level models for different cells.
15. The method of claim 1 , wherein different predictive cell-level models for adjacent cells are based on historical context data of each of the adjacent cells.
16. The method of claim 1 , wherein scheduling network traffic for the mobile device includes optimizing performance of the mobile device based on an obstruction in an adjacent cell to the first cell.
17. The method of claim 1 , wherein scheduling network traffic for the mobile device includes dynamically adjusting resource allocation based on arrival of requests from the plurality of mobile devices.
18. The method of claim 1 , wherein different predictive cell-level models for different cells are verified using real-time network traffic.
19. The method of claim 1 , wherein network parameters are updated for different predictive cell-level models for different cells.
20. The method of claim 1 , wherein the predictive multi-cell model provides for joint decisions to be made by multiple base stations across multiple cells in the mobile network.