Intelligent Caching
▶ Introduction
 

According to the Cisco Visual Networking Index, watching videos from devices (i.e., smart phone, personal computer, smart television) has been generating most of the Internet traffic and is forecast to continue to increase exponentially [1]. In order to handle the overwhelming Internet traffic, several future Internet network architectures (such as Information-Centric Networking, Multi-access Edge Computing) have been proposed with caching capability [2]. Thus, with caching and computing capability, nodes (i.e., base stations, small-cell base stations, access points, routers) can temporarily store popular videos in their cache storage to satisfy user requests in the future, instead of retrieving video from content servers. So, the video contents’ access delay (or) the number of retrieving videos from the content servers is reduced with content caching schemes compared to that without caching schemes.

Caching can be classified into two major categories: i) reactive caching and ii) proactive caching. In reactive caching, the node makes the cache decision (whether to store the requested content or not) only when a request for a particular content arrives [3]. In proactive caching, the node proactively predicts a content’s popularity before any user requests are received and makes the cache decision based on this [4], [5]. In practice, the assumption that the content’s popularity follows some probability distribution may be invalid because the popularity of the content dynamically changes depending on different factors (e.g., events, type of content, and lifespan of the content). Therefore, machine learning becomes major player to predict the content’s popularity as well as to manage the cache storage.

There are three major types of learning process could be utilized: i) supervised learning, ii) unsupervised learning, and iii) reinforcement learning. In supervised learning, we know the correct answer (labeled training data) before training our model, which allows us to make predictions about unseen data. In unsupervised learning, we are dealing with unlabeled data or data of an unknown structure. Unsupervised learning techniques extract meaningful information without the guidance of a known outcome variable. In reinforcement learning, we have a sequential decision-making problem, where making a decision influences what decisions can be made in the future. A reward function is provided and it tells us how ‘‘good’’ certain states are. Currently, a lot of researcher are working on various machine learning based caching schemes such as social community aware caching [6], unmanned aerial vehicle (UAV) assisted caching [7], Virtual Reality/Augmented Reality related caching, etc.

Fig. 1. System model

▶ Research Challenges and Issues

  The most important challenges in machine learning based caching are as follows:
  • It requires high computation resources to process the big data (high-dimensional data) to train the prediction models.
  • It is difficult to find a suitable prediction model among the various types of deep learning models, such as Multilayer Perceptrons (MLPs), Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Convolutional Recurrent Neural Networks (CRNNs), etc. [27].
  • It is difficult to tune parameters such as the number of layers (i.e., the depth of the network), types of layers (e.g., Convolutional, Recurrent and Fully Connected layers), and learning rate to improve the accuracy of the prediction model.


  • ▶ References

     
    1. CISCO VNI. Accessed: Feb. 7, 2019. [Online]. Available: http://www. cisco.com/c/en/us/solutions/service-provider/visual-networking-indexvni/index.html
    2. M. Zhang, H. Luo, and H. Zhang, ‘‘A survey of caching mechanisms in information-centric networking,’’ IEEE Commun. Surveys Tuts., vol. 17, no. 3, pp. 1473?1499, 3rd Quart., 2015.
    3. S. Ullah, K. Thar, and C. S. Hong, ‘‘Management of scalable video streaming in information centric networking,’’ Multimedia Tools Appl., vol. 76, no. 20, pp. 21519?21546, Oct. 2017.
    4. W.-X. Liu, J. Zhang, Z.-W. Liang, L.-X. Peng, and J. Cai, ‘‘Content popularity prediction and caching for ICN: A deep learning approach with SDN,’’ IEEE Access, vol. 6, pp. 5075?5089, 2018, doi: 10.1109/ ACCESS.2017.2781716.
    5. M. Chen, W. Saad, C. Yin, and M. Debbah, ‘‘Echo state networks for proactive caching in cloud-based radio access networks with mobile users,’’ IEEE Trans. Wireless Commun., vol. 16, no. 6, pp. 3520?3535, Jun. 2017, doi: 10.1109/TWC.2017.2683482.
    6. Han Yeo Reum Im, Kyi Thar, and Choong Seon Hong, “Q-Learning Based Social Community-Aware Energy Efficient Cooperative Caching in 5G Networks,” The 11th International Conference on Ubiquitous and Future Networks(ICUFN 2019), July 2 - 5, 2019, Zagreb, Croatia
    7. Seok-Won Kang, Kyi Thar, Choong Seon Hong, "Unmanned Aerial Vehicle Allocation and Deep Learning Based Content Caching in Wireless Network," The International Conference on Information Networking (ICOIN 2020), January 7-10, 2020, Barcelona, Spain


    ▶ Achievements

     
    1. Seok-Won Kang, Kyi Thar, Choong Seon Hong, "Unmanned Aerial Vehicle Allocation and Deep Learning Based Content Caching in Wireless Network," The International Conference on Information Networking (ICOIN 2020), January 7-10, 2020, Barcelona, Spain
    2. Han Yeo Reum Im, Kyi Thar, and Choong Seon Hong, “Q-Learning Based Social Community-Aware Energy Efficient Cooperative Caching in 5G Networks,” The 11th International Conference on Ubiquitous and Future Networks(ICUFN 2019), July 2 - 5, 2019, Zagreb, Croatia
    3. Anselme Ndikumana, Nguyen H. Tran, Tai Manh Ho, Zhu Han, Walid Saad, Dusit Niyato, Choong Seon Hong , "Joint Communication, Computation, Caching, and Control in Big Data Multi-access Edge Computing," IEEE Transactions on Mobile Computing, Vol.19, Issue 6, pp.1359-1374, Jun. 2020
    4. Kyi Thar, Nguyen H. Tran, Thant Zin Oo, Choong Seon Hong, "DeepMEC: Mobile Edge Caching Using Deep Learning," IEEE Access, Vol.6, Issue 1, pp.78260-78275, December 2018
    5. Kyi Thar, Thant Zin Oo, Yan Kyan Tun, Do Hyeon Kim, Ki Tae Kim and Choong Seon Hong, "A Deep Learning Model Generation Framework for Virtualized Multi-access Edge Cache Management," IEEE Access, Vol.7, pp.62734-62749, May. 2019