Title Page
Contents
ABSTRACT 10
국문초록 12
CHAPTER 1. INTRODUCTION 14
CHAPTER 2. RELATED WORK 18
2.1. Centralized Service Caching 18
2.2. Distributed Service Caching 19
CHAPTER 3. SYSTEM MODEL 21
3.1. System architecture 21
3.2. Problem Formulation 23
CHAPTER 4. PROPOSED SERVICE CACHING FRAMEWORK 26
4.1. Overview 26
4.2. FL-based Service Popularity Score 27
4.2.1. Distributed Training Phase 27
4.2.2. Distributed Inference Phase 35
4.3. Service Caching Policies 36
CHAPTER 5. RESULTS 41
5.1. Experimental setup 41
5.2. Metric 44
5.3. Counterpart 45
5.4. Performance Evaluation 46
5.4.1. Comparison of Service hit ratio 46
5.4.2. Comparison of Service delay 50
CHAPTER 6. CONCLUSION 55
REFERENCES 56
Table 1. Summary of Notations 39
Table 2. Simulation parameters. 43
[Figure 3-1] MEC system architecture. 21
[Figure 4-1] Workflow of SCFL of each MEC server. 26
[Figure 4-2] FL-based Service Caching Training Phase. 28
[Figure 4-3] Process of data preparation on each MEC server at time slot tm.[이미지참조] 29
[Figure 4-4] Adversarial Autoencoder Architecture. 31
[Figure 5-1] Service hit ratio with different computational capacity of each MEC server. 46
[Figure 5-2] Service hit ratio with different number of users. 48
[Figure 5-3] Service hit ratio with different number of MEC servers. 49
[Figure 5-4] Service delay with different computational capacity of each MEC server. 50
[Figure 5-5] Service delay with different number of users. 51
[Figure 5-6] Total delay with different number of MEC servers. 53