Page 79 - Proceedings of the 2018 ITU Kaleidoscope

Machine learning for a 5G future

[3] Diehl-Fleig, E., and A. M. De Araujo. "Haplometrosis
and pleometrosis in the antAcromyrmex striatus
(Hymenoptera: Formicidae)." Insectes sociaux 43,
no. 1 (1996): 47-51- Springer
[4] Theraulaz, Guy, Jacques Gautrais, Scott Camazine, &
Jean-Louis Deneubourg. "The formation of spatial
patterns in social insects: from simple behaviours to
complex structures." A Philosophical Transactions of
the Royal Society of London: Mathematical, Physical
and Engineering Sciences 361, no. 1807 (2003):
1263-1282.
[5] Marco Dorigo, Gianni Di Caro and Luca M.
Gambardella, “Ant Algorithms for Discrete
Optimization”, Artificial Life, MIT Press, 1999
[6] Lixiang Li, Haipeng Peng, Jürgen Kurths, Yixian
Yang, and Hans Joachim Schellnhuber, “Chaos–order
Figure 8: Smart combination of different algorithms
transition in foraging behavior of ants”, Proc National
Academy Sciences U.S.A. 2014 Jun 10; 111(23):
classification improves the perception before the reasoning 8392–8397.
is applied to predict the actual non-zero FER value. [7] Stephen K. Reed, “Cognition: Theory and
Application”, 9th, Cengage Learning, 2012
[8] Kihlstrom, John F. "The psychological
These results show that learning methods can be used to
automate network management functions not only to learn unconscious." Handbook of personality. Theory and
the best configurations for specific states but to also learn the research (1990): 424-442.
underlying network response which can then be used to [9] 3rd Generation Partnership Project; Technical
predict response in unknown states. Specification Group Services and System Aspects;
“Telecommunication management; Principles and
high-level requirements (Release 15),” 3GPP, Tech.
5. CONCLUSION AND OUTLOOK
Spec. 32.101 V15.0.0, September 2017. [Online].
Available: http://www.3gpp.org
This paper has outlined a framework for Cognitive Autono- [10] Huaning Niu, Clara (Qian) Li, Apostolos
mous Networks (CAN). A cognitive entity is one capable of
perceiving stimuli, transforming them into data elements Papathanassiou, Geng Wu, “RAN Architecture
Options and Performance for 5G Network Evolution”,
over which it reasons to select actions. It conceptualizes and
contextualizes a data element and logically or arithmetically IEEE WCNC 2014 - Workshop on Wireless
Evolution Beyond 2020
relates it with other data elements to make inferences about
the elements and their relations and, subsequently, to select [11] Mwanje, Stephen S., and Andreas Mitschele-Thiel,
"Distributed cooperative Q-learning for mobility-
the appropriate action. Networks with this capability and the
ability to independently act become CANs. We have sensitive handover optimization in LTE SON."
Computers and Communication (ISCC), 2014 IEEE
proposed a cognition model based on this perception-
Symposium on. IEEE, 2014.
reasoning flow and used it to describe the path towards CAN [12] S. Mwanje, L.C. Schmelz, and A. Mitschele-Thiel,
and a functional design of a typical CAN system for which
AI/ML techniques form a part of each function. We then “Cognitive Cellular Networks: A Q-Learning
Framework for Self-Organizing Networks,” IEEE
showed how this kind of learning can be beneficial in net-
work operations using a use case of learning a network’s Transactions on Network and Service Management,
vol 13, no. 1, pp. 85 – 98, January 2016.
response to different mobility states and handover configu-
rations. Our future work is to build and test a complete CAN [13] Altman, N. S. "An introduction to kernel and nearest-
neighbor nonparametric regression". The American
based on the proposed framework.
Statistician. 46 (3) (1992): 175–
185. doi:10.1080/00031305.1992.10475879.
REFERENCES
[14] T. Hastie, R. Tibshirani & J. Friedman, “Elements of
[1] ITU T. Recommendation, "M. 3400: TMN Statistical Learning”, Springer, 2009. And L. Breinan
management functions." (2000). [15] Random Forests regressor - L. Breiman, “Random
[2] S. Hämäläinen, H. Sanneck, C. Sartori (Eds.), “LTE Forests”, Machine Learning, 45(1), 5-32, 2001.
Self-Organising Networks (SON),” John Wiley & [16] P. Geurts, D. Ernst., and L. Wehenkel, “Extremely
Sons, 2012 randomized trees”, Machine Learning, 63(1), 3-42,
2006.

– 63 –