Loading...

Partial Coded Data Placement in Cache Memories in Information Centric Networks

Salehi, Mohammad Javad | 2018

411 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 51450 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hossein Khalaj, Babak; Motahari, Abolfazl
  7. Abstract:
  8. In this research, we have studied the effect of coding techniques, as well as partial data placement in cache memories throughout the network. Cache memories play an important role in information centric networks. As content delivery account for the major part of data transfers in current IP-based data networks, these network structures have gained much attention recently and hence optimizing cache performance in their structure is of great importance.Throughout this research, we first analyze the effect of simultaneous data delivery from multiple cache memories on the network performance. We consider two basic models for the network graph; and show there exists a fundamental trade-off among three main parameters being considered: observed delay, request service cost and the minimum required cache size. We analyze this trade-off mathematically and find its governing equations for the considered network model.Next, we enhance the considered model and study the effect of partial data placement in cache networks on the total average service cost. We show two essential laws exist for optimal data placement in cache networks. The first law states that under simple conditions, the optimal data placement in cache memories throughout the network is non-redundant necessarily. The second law, named 0-1 law, states that for every non-redundant non-0-1 data placement in cache networks, there exists at least one non-redundant 0-1 data placement for which the total average service cost is not increased. Combining the two aforementioned laws, we find conditions for which the optimal cache placement in the network is both non-redundant and 0-1.The results of this research provide a good insight on the possible improvements in the network performance, in the presence of coding techniques and partial data placement in cache memories. Furthermore, these results can be used in future, for designing novel cache placement algorithms, which provide acceptable performance, operational complexity and execution speed at the same time. One such algorithm is presented in this research and its performance is analyzed through extensive simulations
  9. Keywords:
  10. Information Centric Networks ; Cache Memory ; Partial Data Placement ; Coded Data Placement

 Digital Object List

 Bookmark

No TOC