ZTE Communications ›› 2019, Vol. 17 ›› Issue (2): 10-18.DOI: 10.12142/ZTECOM.201902003
• Special Topic • Previous Articles Next Articles
YE Dezhong1, LV Haibing1, GAO Yun2, BAO Qiuxia2, CHEN Mingzi2
Received:
2018-09-28
Online:
2019-06-11
Published:
2019-11-14
About author:
YE Dezhong received his B.S. degree from Jilin University, China in 1996. He is currently a chief big data engineer at ZTE Corporation. His research interests include big data in wireline communication and fixed network product|LV Haibing received his B.S. degree from China University of Mining and Technology, China in 2002. He is currently a wireline product architecture director at ZTE Corporation. His research interests include fixed network products and IPTV systems|GAO Yun received his B.E. degree from Nanjing University of Posts and Telecommunications (NUPT), China in 2016. He is currently pursing Ph.D. degree in NUPT. His research interests include machine learning, deep learning, and QoE in multimedia communication|BAO Qiuxia received her B.E. degree from Nanjing University of Posts and Telecommunications (NUPT), China in 2017. She is currently pursing M.S. degree in NUPT. Her research interests include machine learning in big data, etc|Chen Mingzi (Supported by:
YE Dezhong, LV Haibing, GAO Yun, BAO Qiuxia, CHEN Mingzi. Novel Real-Time System for Traffic Flow Classification and Prediction[J]. ZTE Communications, 2019, 17(2): 10-18.
Add to citation manager EndNote|Ris|BibTeX
URL: https://zte.magtechjournal.com/EN/10.12142/ZTECOM.201902003
Indicators | Meanings |
---|---|
Noid | ID of the base station |
Name | Location information of the base station |
Time | Collecting time of the data |
kb | Traffic value of the base station by one day |
Area | Administrative region of the base station |
Table 1 Data attributes
Indicators | Meanings |
---|---|
Noid | ID of the base station |
Name | Location information of the base station |
Time | Collecting time of the data |
kb | Traffic value of the base station by one day |
Area | Administrative region of the base station |
Ratings | Percentage |
---|---|
0-5 | 57% |
5-10 | 15% |
10-15 | 12% |
more than 15 | 11% |
Table 2 Distributions of ratings for dynamic time warping (DTW)
Ratings | Percentage |
---|---|
0-5 | 57% |
5-10 | 15% |
10-15 | 12% |
more than 15 | 11% |
Algorithms | NRMSE |
---|---|
XGBoost | 0.012 |
Linear regression | 0.089 |
ARIMA | 0.017 |
MLP | 0.050 |
Wavelet transform | 0.157 |
LSTM | 0.062 |
Table 3 Results of NRMSE for six algorithms
Algorithms | NRMSE |
---|---|
XGBoost | 0.012 |
Linear regression | 0.089 |
ARIMA | 0.017 |
MLP | 0.050 |
Wavelet transform | 0.157 |
LSTM | 0.062 |
R2 | Integrated model | ARMA | WT | LSTM |
---|---|---|---|---|
≥0.8 | 8.5% | 6.11% | 7.50% | 7.46% |
≥0.5 | 19.46% | 15.2% | 18.4% | 17.39% |
≥0 | 59.32% | 41.0% | 59.0% | 59.1% |
Table 5 Results of R2 in the Experiments
R2 | Integrated model | ARMA | WT | LSTM |
---|---|---|---|---|
≥0.8 | 8.5% | 6.11% | 7.50% | 7.46% |
≥0.5 | 19.46% | 15.2% | 18.4% | 17.39% |
≥0 | 59.32% | 41.0% | 59.0% | 59.1% |
MAE | Integrated model | ARMA | WT | LSTM |
---|---|---|---|---|
≤15% | 60.12% | 22.9% | 51.3% | 56.6% |
≤25% | 81.14% | 35.4% | 70.5% | 75.1% |
≤30% | 86.06% | 52.6% | 77.4% | 82.5% |
≤50% | 93.23% | 74.6% | 86.5% | 90.0% |
≤1 | 96.98% | 95.0% | 95.3% | 96.0% |
Table 4 Results of Normalized Root Mean Square Error (NRMSE) in the Experiments
MAE | Integrated model | ARMA | WT | LSTM |
---|---|---|---|---|
≤15% | 60.12% | 22.9% | 51.3% | 56.6% |
≤25% | 81.14% | 35.4% | 70.5% | 75.1% |
≤30% | 86.06% | 52.6% | 77.4% | 82.5% |
≤50% | 93.23% | 74.6% | 86.5% | 90.0% |
≤1 | 96.98% | 95.0% | 95.3% | 96.0% |
Features | Meanings |
---|---|
1 | Average values one week before |
2 | Average values two weeks before |
3 | Average values three weeks before |
Week_day | Today is a weekend or working day |
Date_int | The concrete date of today |
Table 6 Feature selection in the experiment
Features | Meanings |
---|---|
1 | Average values one week before |
2 | Average values two weeks before |
3 | Average values three weeks before |
Week_day | Today is a weekend or working day |
Date_int | The concrete date of today |
[1] | ZHOU L, WU D, DONG Z J ,et al.When Collaboration Hugs Intelligence: Content Delivery over Ultra-Dense Networks[J]. IEEE Communications Magazine, 2017,55(12):91-95. DOI: 10.1109/mcom.2017.1700481 |
[2] | ZHOU L, WU D, CHEN J X ,et al.Greening the Smart Cities: Energy-Efficient Massive Content Delivery via D2D Communications[J]. IEEE Transactions on Industrial Informatics, 2018,14(4):1626-1634. DOI: 10.1109/tii.2017.2784100 |
[3] | WU D, WANG J L, HU R Q ,et al.Energy-Efficient Resource Sharing for Mobile Device-To-Device Multimedia Communications[J]. IEEE Transactions on Vehicular Technology, 2014,63(5):2093-2103. DOI: 10.1109/tvt.2014.2311580 |
[4] | ZHAO T S, LIU Q, CHEN C W . QoE in Video Transmission: A User Experience-Driven Strategy[J]. IEEE Communications Surveys & Tutorials, 2017,19(1):285-302. DOI: 10.1109/comst.2016.2619982 |
[5] | PAMULA T . Classification and Prediction of Traffic Flow Based on Real Data Using Neural Networks[J]. Archives of Transport, 2012,24(4):519-529. DOI: 10.2478/v10174-012-0032-2 |
[6] | GUO F C, POLAK J W, KRISHNAN R. Comparison of Modelling Approaches for Short Term Traffic Prediction under Normal and Abnormal Conditions [C]//13th International IEEE Conference on Intelligent Transportation Systems. Funchal, Portugal, 2010: 1209-1214. DOI: 10.1109/ITSC.2010.5625291 |
[7] | WILLIAMS N, ZANDER S, ARMITAGE G . A Preliminary Performance Comparison of Five Machine Learning Algorithms for Practical IP Traffic Flow Classification[J]. ACM SIGCOMM Computer Communication Review, 2006,36(5):5. DOI: 10.1145/1163593.1163596 |
[8] | HU C C, YAN X H. Mining Traffic Flow Data Based on Fuzzy Clustering Method [C]//Fourth International Workshop on Advanced Computational Intelligence. Wuhan, China, 2011: 245-248. DOI: 10.1109/IWACI.2011.6160011 |
[9] | HÖCHST J, BAUMGÄRTNER L, HOLLICK M, et al. Unsupervised Traffic Flow Classification Using a Neural Autoencoder [C]//IEEE 42nd Conference on Local Computer Networks (LCN). Singapore, Singapore, 2017: 523-526. DOI: 10.1109/LCN.2017.57 |
[10] | JEONG Y S, JEONG M K, OMITAOMU O A . Weighted Dynamic Time Warping for Time Series Classification[J]. Pattern Recognition, 2011,44(9):2231-2240. DOI: 10.1016/j.patcog.2010.09.022 |
[11] | CHEN T, GUESTRIN C,. XGBoost: A Scalable Tree Boosting System [C]//ACM International Conference on Knowledge Dicovery and Data Mining. San Francisco, USA, 2016: 785-794 |
[12] | LI Z, LEI Q, XUE K Y, et al. A Novel BP Neural Network Model for Traffic Prediction of Next Generation Network [C]//Fifth International Conference on Natural Computation, Tianjin, China, 2009: 32-38. DOI: 10.1109/ICNC.2009.673 |
[13] | YANG W, YANG D Y, ZHAO Y L, et al. Traffic Flow Prediction Based on Wavelet Transform and Radial Basis Function Network [C]//International Conference on Logistics Systems and Intelligent Management (ICLSIM). Harbin, China, 2010: 969-972. DOI: 10.1109/ICLSIM.2010.5461098 |
[14] | ZAHARIA M, CHOWDHURY M, FRANKLIN M, et al. Spark: Cluster Computing with Working Sets [C]//2nd Usenix Conference on Hot Topicsin Cloud Computing. Boston, USA, 2010: 10-10 |
[15] | DEAN J, GHEMAWAT S. MapReduce: Simplified Data Processing on Large Clusters [C]//6th Symposium on Operating Systems Design & Implementaion. San Francisco, USA, 2004: 137-149 |
[16] | SAKOE H, CHIBA S . Dynamic Programming Algorithm Optimization for Spoken Word Recognition[J]. IEEE Transactions on Acoustics, Speech, and Signal Processing, 1978,26(1):43-49. DOI: 10.1109/tassp.1978.1163055 |
[17] | MA X L, DING C, LUAN S ,et al. Prioritizing Influential Factors for Freeway Incident Clearance Time Prediction Using the Gradient Boosting Decision Trees Method[J]. IEEE Transactions on Intelligent Transportation Systems, 2017,18(9):2303-2310. DOI: 10.1109/tits.2016.2635719 |
No related articles found! |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||