1 / 58

Inferring Mobile Traffic Patterns with ZipNet-GAN

This study presents ZipNet-GAN, a Generative Adversarial Neural Network, for inferring fine-grained mobile traffic patterns. It examines social analysis, mobility analysis, and network analysis to gain insights into traffic and social features, users' interactions and demographic research, traffic and users' mobility, movement prediction, transportation planning, and traffic and network conditions, resource allocation, and fault diagnosis.

bmyrick
Télécharger la présentation

Inferring Mobile Traffic Patterns with ZipNet-GAN

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. ZipNet-GAN: Inferring Fine-grained Mobile Traffic Patterns via a Generative Adversarial Neural Network Chaoyun Zhang, Xi Ouyang, and Paul Patras Huazhong University of Science and Technology University of Edinburgh

  2. The Goal of Mobile Traffic Analysis The Goal of Mobile Traffic Analysis 1. Social analysis -- Traffic and social features • User's interactions • Demographic research Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  3. The Goal of Mobile Traffic Analysis The Goal of Mobile Traffic Analysis 1. Social analysis -- Traffic and social features • User's interactions • Demographic research 2. Mobility analysis-- Traffic and users' mobility • Movement prediction • Transportation planning Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  4. The Goal of Mobile Traffic Analysis The Goal of Mobile Traffic Analysis 1. Social analysis -- Traffic and social features • User's interactions • Demographic research 2. Mobility analysis-- Traffic and users' mobility • Movement prediction • Transportation planning 3. Network analysis -- Traffic and network condition • Resource allocation • Fault diagnosis Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  5. The Goal of Mobile Traffic Analysis The Goal of Mobile Traffic Analysis Require fine-grained knowledge of mobile traffic !!! 1. Social analysis -- Traffic and social features • User's interactions • Demographic research 2. Mobility analysis-- Traffic and users' mobility • Movement prediction • Transportation planning 3. Network analysis -- Traffic and network condition • Resource allocation • Fault diagnosis Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  6. Fine-Grained Traffic Measurement Fine-Grained Traffic Measurement 1. Relies on dedicated probes • Packet Gateway (PGW) probes -- yield approximated position Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  7. Fine-Grained Traffic Measurement Fine-Grained Traffic Measurement 1. Relies on dedicated probes • Packet Gateway (PGW) probes -- yield approximated position • Radio Network Controller (RNC) probes -- have small geographical coverage Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  8. Fine-Grained Traffic Measurement Fine-Grained Traffic Measurement 1. Relies on dedicated probes , • Packet Gateway (PGW) probes -- yield approximated position • Radio Network Controller (RNC) probes -- have small geographical coverage 2. Involves data post-processing overhead • Call detail record reports transfer • Data storage • Spatial aggregation Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  9. Coarse-Grained to Fine-Grained Coarse-Grained to Fine-Grained Fine-grain traffic measurement (desired) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  10. Coarse-Grained to Fine-Grained Coarse-Grained to Fine-Grained Aggregation Coarse-grain traffic measurement (Aggregated) Fine-grain traffic measurement (desired) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  11. Coarse-Grained to Fine-Grained Coarse-Grained to Fine-Grained ? Costly! Coarse-grain traffic measurement (Aggregated) Fine-grain traffic measurement (desired) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  12. Coarse-Grained to Fine-Grained Coarse-Grained to Fine-Grained ? Machine Learning? Coarse-grain traffic measurement (Aggregated) Fine-grain traffic measurement (desired) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  13. Mobile Traffic Super-Resolution (MTSR) * Image source: https://www.pinterest.com/ Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  14. Mobile Traffic Super-Resolution (MTSR) Mobile Traffic Super-Resolution (MTSR) * Image source: https://www.pinterest.com/ Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  15. Mobile Traffic Super-Resolution (MTSR) Mobile Traffic Super-Resolution (MTSR) Both learn the correlation between low-resolution and high-resolution “frames”! * Image source: https://www.pinterest.com/ Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  16. Mobile Traffic Super-Resolution (MTSR) Mobile Traffic Super-Resolution (MTSR) Inspiration: Apply image SR techniques to the MTSR ! * Image source: https://www.pinterest.com/ Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  17. Solution: Generative Adversarial Nets (GANs) 1. An unsupervised deep learning framework. S S Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  18. Solution: Generative Adversarial Nets (GANs) Generative Adversarial Nets (GANs) 1. An unsupervised deep learning framework. 2. Learn the target distribution and generate artificial data. S S Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  19. Solution: Generative Adversarial Nets (GANs) Generative Adversarial Nets (GANs) 1. An unsupervised deep learning framework. 2. Learn the target distribution and generate artificial data. 3. Force the output of the model to be closer to real data distributions. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  20. ZipNet-GAN ZipNet-GAN The Generator (ZipNet) The Discriminator (VGG-net) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  21. ZipNet-GAN ZipNet-GAN The Generator (ZipNet) The Discriminator (VGG-net) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  22. ZipNet-GAN ZipNet-GAN The Generator (ZipNet) The Discriminator (VGG-net) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  23. ZipNet-GAN ZipNet-GAN The Generator (ZipNet) The Discriminator (VGG-net) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  24. ZipNet-GAN ZipNet-GAN The Generator (ZipNet) The Discriminator (VGG-net) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  25. Dataset Dataset • Released by Telecom Italia's Big Data Challenge Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  26. Dataset Dataset • Released by Telecom Italia's Big Data Challenge • Measurement of mobile traffic volume between 1 Nov 2013 and 1 Jan 2014 in Milan. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  27. Dataset Dataset • Released by Telecom Italia's Big Data Challenge • Measurement of mobile traffic volume between 1 Nov 2013 and 1 Jan 2014 in Milan. • Aggregated every 10 minutes. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  28. Dataset Dataset • Released by Telecom Italia's Big Data Challenge • Measurement of mobile traffic volume between 1 Nov 2013 and 1 Jan 2014 in Milan. • Aggregated every 10 minutes. • Partitioned in 100×100 grids. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  29. Four MTSR Instances Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  30. Four MTSR Instances Four MTSR Instances Green: 10×10 Probes Yellow: 4×4 Probes Red: 2×2 Probes Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  31. Methods for Comparison Methods for Comparison • Interpolations • Uniform Interpolation (Used by operators) • Bicubic Interpolation Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  32. Methods for Comparison Methods for Comparison • Interpolations • Uniform Interpolation (Used by operators) • Bicubic Interpolation 2. Image Super-Resolution • Sparse Coding (SC) • Adjusted Anchored Neighborhood Regression (A+) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  33. Methods for Comparison Methods for Comparison • Interpolations • Uniform Interpolation (Used by operators) • Bicubic Interpolation 2. Image Super-Resolution • Sparse Coding (SC) • Adjusted Anchored Neighborhood Regression (A+) 3. Deep Learning • Super-Resolution Convolutional Neural Network (SRCNN) • ZipNet • ZipNet-GAN Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  34. Methods for Comparison Methods for Comparison Train on data collected in 40 days, validate on 10 days, test on 10 days. • Interpolations • Uniform Interpolation (Used by operators) • Bicubic Interpolation 2. Image Super-Resolution • Sparse Coding (SC) • Adjusted Anchored Neighborhood Regression (A+) 3. Deep Learning • Super-Resolution Convolutional Neural Network (SRCNN) • ZipNet • ZipNet-GAN Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  35. Numerical Results Results Normalised Root Mean Square Error: The prediction accuracy. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  36. Numerical Results Results Normalised Root Mean Square Error: The prediction accuracy. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  37. Numerical Results Results Normalised Root Mean Square Error: The prediction accuracy. Peak Signal-to-Noise Ratio: The quality of reconstruction. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  38. Numerical Results Results Normalised Root Mean Square Error: The prediction accuracy. Peak Signal-to-Noise Ratio: The quality of reconstruction. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  39. Numerical Results Results Normalised Root Mean Square Error: The prediction accuracy Peak Signal-to-Noise Ratio: The quality of reconstruction. Structural Similarity: The similarity between frames. Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  40. Numerical Results Results Enhance the resolution of mobile traffic measurements by 100×. 78% lower reconstruction error, 40% higher fidelity of reconstruction patterns, 36.4× structural similarity! Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  41. MTSR Examples (Up-10) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  42. MTSR Examples (Up-10) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  43. MTSR Examples (Up-10) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  44. MTSR Examples (Up-10) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  45. MTSR Examples (Up-10) Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  46. Impacts of the GAN Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  47. Operating with Anomalous Traffic Working with Anomaly Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  48. Operating with Anomalous Traffic Working with Anomaly Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  49. Operating with Anomalous Traffic Working with Anomaly Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

  50. Operating with Anomalous Traffic Working with Anomaly Can operate under anomalous traffic situations ! Chaoyun Zhang et al. ZipNet-GAN... December 01, 2017 Chaoyun Zhang et al. ZipNet-GAN... December 14, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 16, 2017 Chaoyun Zhang et al. ZipNet-GAN... November 30, 2017

More Related