Performance Analysis of CDN Caching Strategies for Global Applications

Abstract:

Content Delivery Networks (CDNs) reduce latency and cloud egress costs for global applications; however, cache efficiency varies significantly with configuration choices [1]. Most production CDNs employ origin-pull caching, which minimizes storage requirements and operational complexity, but cache hit ratios and performance depend heavily on cache size, eviction policies, and content characteristics [2].

This project uses NS-3 network simulation to optimize origin-pull CDN configurations under realistic global traffic patterns. The simulation models a production application with origin servers in Azure East US serving users across North America (20ms baseline), Europe (80ms), Asia (180ms), and Australia (220ms), with edge nodes strategically placed in 8 geographic regions.

The research evaluates cache configurations ranging from 10GB to 500GB per edge node using three eviction policies: LRU (Least Recently Used), LFU (Least Frequently Used), and TTL-based expiration [5]. Content distribution reflects typical web applications: static assets (40%, highly cacheable), API responses (35%, conditionally cacheable with 5-minute TTL), HTML pages (20%, short TTL), and user-specific content (5%, non-cacheable).

Performance metrics include cache hit ratio by content type, Time-To-First-Byte (TTFB) improvement by geographic region, origin bandwidth reduction percentage, and storage efficiency across cache sizes [3]. Test scenarios examine steady-state traffic patterns, flash events (50x traffic spike for 2 hours simulating product launches), global cache invalidation during deployments, regional failover behaviour, and cold-start performance for new edge locations [4].

The research addresses practical questions such as: "What cache hit ratio justifies edge node deployment costs?" and "Does increasing cache size from 100GB to 500GB provide meaningful performance gains?" Expected results will quantify the relationship among cache configuration, performance improvements, and infrastructure costs [6], thereby enabling evidence-based decisions on CDN architecture to optimize global application delivery.

References:
[1] E. Nygren, R. K. Sitaraman, and J. Sun, "The Akamai Network: A Platform for High-Performance Internet Applications," ACM SIGOPS Operating Systems Review, vol. 44, no. 3, pp. 2-19, Jul. 2010.
[2] J. Dilley, B. Maggs, J. Parikh, H. Prokop, R. Sitaraman, and B. Weihl, "Globally Distributed Content Delivery," IEEE Internet Computing, vol. 6, no. 5, pp. 50-58, Sep./Oct. 2002.
[3] S. Triukose, Z. Wen, and M. Rabinovich, "Measuring a Commercial Content Delivery Network," in Proc. 20th International Conference on World Wide Web (WWW), 2011, pp. 467-476.
[4] V. K. Adhikari, Y. Guo, F. Hao, M. Varvello, V. Hilt, M. Steiner, and Z. L. Zhang, "Unreeling Netflix: Understanding and Improving Multi-CDN Movie Delivery," in Proc. IEEE INFOCOM, 2012, pp. 1620-1628.
[5] S. Podlipnig and L. Boszormenyi, "A Survey of Web Cache Replacement Strategies," ACM Computing Surveys, vol. 35, no. 4, pp. 374-398, Dec. 2003.
[6] D. S. Berger, P. Gland, S. Singla, and F. Cidon, "Exact Analysis of TTL Cache Networks," in Proc. ACM SIGMETRICS Performance Evaluation Review, vol. 42, no. 1, pp. 389-400, Jun. 2014.

Abstract:

References:

Machine-Learning Analysis of Recent Internet Outages Using IODA Data

Abstract:
The Internet Outage Detection and Analysis (IODA) platform provides multi-signal time-series data to monitor large-scale Internet disruptions. Prior work has shown that sliding-window machine-learning methods can detect outage events in selected case studies, but generalization to new outage scenarios remains an open question.

This project analyzes recent IODA outage events using raw time-series data and evaluates multiple machine-learning models under a consistent sliding-window framework. The goal is to assess robustness across different outage types and identify practical approaches for reliable anomaly detection.

References:
[1] Internet Outage Detection and Analysis (IODA), Georgia Tech Internet Intelligence Lab.
[2] L. F. Oliveira et al., "Internet outages during times of conflict," Proc. IEEE SMC, 2024.
[3] Z. Li and L. Trajkovic, "Machine learning for detecting anomalies in communication networks," IEEE JSAC.

Performance Analysis of VR Streaming Under Wi-Fi 6

Abstract:
Virtual reality (VR) gaming refers to running video games on a computer while using a VR headset as the display, which results in a more immersive experience than a monitor. Due to the immersive experience, the user base is expanding [1]. Now VR headset manufactures supports wireless streaming, which imposes a high requirement on both bandwidth (sending the images from the computer to the headset) and latency (sending the user input from the headset to the computer) of the network [2]. However, building a dedicated network for VR streaming [LH1.1]is expensive and may be wasteful. Therefore, to ensure a good quality of experience (QoE) [3] in a shared network, this project first assumes one user is streaming games to their VR headset, while other users are browsing the web or watching videos [4] using the same Wi-Fi 6 network [5], then analyzes how would other users’ traffic affect the VR gaming experience in terms of throughput available to the headset and latency experienced by the headset using ns-3.

References:
[1] Statista, "Virtual reality (VR) software B2C market revenue worldwide from 2017 to 2028, by segment," Nov. 17, 2023. [Online]. Available: https://www.statista.com/forecasts/1337178/vr-software-b2c-market-revenue-worldwide-by-segment/. [Accessed Jan. 10, 2026].
[2] S. Mangiante, K. Guenter, A. Navon, G. Zhuang, R. Ju and M. Silva, "VR is on the Edge: How to Deliver 360° Videos in Mobile Networks," in Proceedings of VR/AR Network '17., Aug. 25, 2017, doi: 10.1145/3097895.3097901.
[3] G. Su, X. Su, Y. Bai, M. Wang, A. Vasilakos and H. Wang,"QoE in video streaming over wireless networks: perspectives and research challenges," Wireless Netw 22., vol. 22, pp. 1571–1593, 2016, doi:10.1007/s11276-015-1028-7.
[4] A. Rao, A. Legout, Y. Lim, D. Towsley, C. Barakat and W. Dabbous, "Network characteristics of video streaming traffic," in CoNEXT '11: Proceedings of the Seventh COnference on emerging Networking EXperiments and Technologies., Dec. 6, 2011, doi: 10.1145/2079296.2079321.
[5] E. Khorov, A. Kiryanov, A. Lyakhov and G. Bianchi, "A Tutorial on IEEE 802.11ax High Efficiency WLANs," in IEEE Communications Surveys & Tutorials, vol. 21, no. 1, pp. 197-216, Firstquarter 2019, doi: 10.1109/COMST.2018.2871099.

AQM Shootout in ns-3: Quantifying Bufferbloat and Latency to Throughput Tradeoffs in Home Networks

Abstract:
Bufferbloat, excessive latency caused by oversized network buffers under load, remains a persistent problem in home and edge networks. This project proposes a systematic evaluation of Active Queue Management AQM algorithms using the ns-3 network simulator to identify configurations that best mitigate delay while preserving throughput and fairness. A dumbbell topology will be used to emulate a residential bottleneck link with realistic propagation delays. Traffic will consist of 1 to N long lived TCP flows competing with a latency sensitive UDP flow representative of gaming or voice applications. The study will compare FQ CoDel, PIE, and RED across parameter sweeps of bottleneck capacity, base RTT, queue discipline, and TCP congestion control variants such as CUBIC, BBR, and DCTCP. Key metrics include 95th percentile queueing delay, end to end latency, aggregate throughput, fairness using Jain’s index, and packet drop or mark rates. A stretch goal will enable Explicit Congestion Notification ECN to contrast marking based and drop based behaviors. The outcome will be a reproducible ns-3 experiment framework, comparative performance plots, and a recommendation matrix mapping optimal AQM choices to traffic mixes and network conditions typical of home environments.

References:
[1] Nichols, K., and Jacobson, V. (2012). Controlling Queue Delay CoDel. ACM Queue, 10(5).
[2] Hoeiland Joergensen, T., McKenney, P., Taht, D., Gettys, J., and Dumazet, E. (2015). The FlowQueue CoDel Packet Scheduler and Active Queue Management Algorithm. RFC 8290, IETF.
[3] Pan, R., Natarajan, P., Baker, F., and VerSteeg, B. (2013). PIE A Lightweight Control Scheme to Address the Bufferbloat Problem. RFC 8033, IETF.
[4] Floyd, S., and Jacobson, V. (1993). Random Early Detection Gateways for Congestion Avoidance. IEEE ACM Transactions on Networking, 1(4), 397 to 413.
[5] ns 3 Consortium. (2023). ns 3 Network Simulator Documentation. https://www.nsnam.org/docs/

QoS Enforcement for Bidirectional UDP-Based Control and Telemetry Traffic Using Packet Scheduling

Abstract:
Many networked control systems rely on the User Datagram Protocol (UDP) to achieve low-latency communication between a centralized controller and multiple devices. These systems typically exchange bidirectional traffic, consisting of controller-to-device control messages and device-to-controller telemetry data, each with different Quality of Service (QoS) requirements. Because UDP provides no built-in congestion control or reliability, such systems are vulnerable to increased delay and packet loss when deployed on shared networks.

This project proposes to evaluate the effectiveness of packet scheduling algorithms in enforcing QoS for bidirectional UDP-based control and telemetry traffic. Using the ns-3 network simulator, a simple client–server topology will be modeled in which a centralized orchestrator communicates with multiple devices over a shared bottleneck link in the presence of background traffic. Packet scheduling disciplines such as First-In First-Out (FIFO), Priority Queuing, and Weighted Fair Queuing will be compared.

Performance metrics including end-to-end delay, jitter, packet loss, and throughput will be used to assess how well each scheduling approach supports the QoS requirements of control and telemetry traffic.

References:
[1] J. Postel, "User Datagram Protocol," RFC 768, Aug. 1980.
[2] A. Demers, S. Keshav, and S. Shenker, "Analysis and simulation of a fair queueing algorithm," ACM SIGCOMM Computer Communication Review, vol. 19, no. 4, pp. 1–12, Sept. 1989.
[3] S. Floyd and V. Jacobson, "Link-sharing and resource management models for packet networks," IEEE/ACM Transactions on Networking, vol. 3, no. 4, pp. 365–386, Aug. 1995.
[4] R. Braden, D. Clark, and S. Shenker, "Integrated services in the Internet architecture: an overview," RFC 1633, June 1994.
[5] ns-3 Consortium, "ns-3 Network Simulator"