Topic summer term 2020: Current Advances in 5G Network Softwarization/Machine Learning techniques
Networks have traditionally been structured with a lot of their functionality set in hardware: most of a router's or a switch's functionality is embedded in custom chipsets. This made them fast but inflexibile. Recently, and particularly with the advent of 5G, the idea of network softwarization has taken hold: Replace hardware-based implementations by software running on commodity hardware. This entails a large number of consequences for the architecture and operation of large-scale networks. This seminar will look into multiple aspects.
Specific topics for summer term 2020
Here is a preliminiary list of possible topics, each with one research papers to start from. To be extended. Each student should pick one paper and extend from there.
Slabicki, Mariusz, Gopika Premsankar, and Mario Di Francesco. "Adaptive configuration of LoRa networks for dense IoT deployments." NOMS 2018-2018 IEEE/IFIP Network Operations and Management Symposium. IEEE, 2018.
Lee, Huang-Chen, and Kai-Hsiang Ke. "Monitoring of large-area IoT sensors using a LoRa wireless mesh network system: Design and evaluation." IEEE Transactions on Instrumentation and Measurement 67.9 (2018): 2177-2187.
El Chall, Rida, Samer Lahoud, and Melhem El Helou. "LoRaWAN network: radio propagation models and performance evaluation in various environments in Lebanon." IEEE Internet of Things Journal 6.2 (2019): 2366-2378.
H. Liang, X. Zhang, J. Zhang, Q. Li, S. Zhou, L. Zhao: "A Novel Adaptive Resource Allocation Model Based on SMDP and Reinforcement Learning Algorithm in Vehicular Cloud System", URL
L. Lei, H. Xu, X. Xiong, K. Zheng, W. Ziang, X. Wang: "Multiuser Resource Control With Deep Reinforcement Learning in IoT Edge Computing", URL
X. Fu, F. R. Yu, J. Wang, Q. Qi: "Dynamic Service Function Chain Embedding for NFV- Enabled IoT: A Deep Reinforcement Learning Approch", URL
S. R. Chowdhury, Anthony, H. Bian, T. Bai and R. Boutaba, “µNF: A Disaggregated Packet Processing Architecture," 2019 IEEE Conference on Network Softwarization (NetSoft), Paris, France, 2019, pp. 342-350. doi: 10.1109/NETSOFT.2019.8806657, URL
J. Santos, T. Wauters, B. Volckaert and F. De Turck, "Towards Network-Aware Resource Provisioning in Kubernetes for Fog Computing Applications," 2019 IEEE Conference on Network Softwarization (NetSoft), Paris, France, 2019, pp. 351-359. doi: 10.1109/NETSOFT.2019.8806671, URL
S. Sharma, N. Uniyal, B. Tola and Y. Jiang, "On Monolithic and Microservice Deployment of Network Functions," 2019 IEEE Conference on Network Softwarization (NetSoft), Paris, France, 2019, pp. 387-395.doi: 10.1109/NETSOFT.2019.8806705, URL
- A Deep Learning Approach to VNF Resource Prediction using Correlation between VNFs
- Unveiling the potential of Graph Neural Networks for network modeling and optimization in SDN
- NeuroViNE: A Neural Preprocessor for Your Virtual Network Embedding Algorithm
E. T. G. D. Sousa, F. A. A. Lins, E. A. G. Tavares and P. R. M. Maciel, "Performance and Cost Modeling Strategy for Cloud Infrastructure Planning," 2014 IEEE 7th International Conference on Cloud Computing, Anchorage, AK, 2014, pp. 546-553. doi: 10.1109/CLOUD.2014.79 URL
Xuan-Qui Pham and Eui-Nam Huh, "Towards task scheduling in a cloud-fog computing system," 2016 18th Asia-Pacific Network Operations and Management Symposium (APNOMS), Kanazawa, 2016, pp. 1-4. doi: 10.1109/APNOMS.2016.7737240 URL
G. Iosifidis, I. Koutsopoulos and G. Smaragdakis, "Distributed Storage Control Algorithms for Dynamic Networks," in IEEE/ACM Transactions on Networking, vol. 25, no. 3, pp. 1359-1372, June 2017. doi: 10.1109/TNET.2016.2633370 URL
AuTO: Scaling Deep Reinforcement Learning for Datacenter-Scale Automatic Traffic Optimization https://conferences.sigcomm.org/events/apnet2018/papers/auto.pdf
Is advance knowledge of flow sizes a plausible assumption?https://www.usenix.org/system/files/nsdi19-dukic.pdf
Learning Scheduling Algorithms for Data Processing Clusters https://web.mit.edu/decima/content/sigcomm-2019.pdf
The goals of a seminar (Master) or proseminar (Bachelor) are to introduce and practice the reading, writing, and presentation of technical and scientific content. This includes, yet is not limited to:
- Independent understanding and production of content based on original literature
- Finding suitable sources based on first hints
- Selecting important content and disregarding less relevant material
- Preparing a writeup (an excellent exercise for later production of Master thesis and similar documents)
- Presenting content to an audience
Seminars and proseminars are structurally very similar; seminars address students in a Master program and hence have slightly higher expectation levels regarding content as well as independence of work.
We typically run seminars as "mini conferences", in a block format. Participants will assumes the roles of "authors" in such a conference as well as those of "reviewers". This will introduce a crucial aspects of the scientific community and its processes; in addition, it will also give participants a broader, more critical understanding of text production and reception.
There are a couple of typical steps:
- Assignment of topics
- VERY short review of assigned literature, identifying a set of sources to work from
- Structure of the writeup
- Writeup in a draft format
- Mutual review of drafts amongst participants
- Final version of writeup
- Draft version of presentation (slides or similar)
- Final version of presentation
- Actual presentation
The grade of a seminar comprises aspects of text production, independence, originality, presentation quality, and discussions during the actual "conference".
Plagiarism is an annoying yet repeating issue in such events. We will extensively discuss what constitutes plagiarism and help to avoid it. But we will also not tolerate any form of plagiarism and strictly follow procedures as specified in the exam regulations.