Loading...

Approximation of Delay-Memory Trade-off for Cache-aided Multiuser Channels

Zamani, Amir Reza | 2018

488 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51123 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Maddah Ali, Mohammad Ali; Mirmohseni, Mahtab
  7. Abstract:
  8. In last decades, network traffic has shifted from voice to content, which can cause network congestion. A common technique to reduce the chance of congestion during peaktraffic periods, which leads to fewer congestion, is to duplicate contents in distributed memories across the network (servers, users, routers, ...). This technique is called caching. We study a cache-aided Gaussian broadcast channel (BC), where a transmitter is connected to two users through a Gaussian BC. The transmitter has access to a server with N equal size files. Each user has a local cache memory which is filled up in the placement phase. In the delivery phase, each user requests one of the files from the transmitter that is responsible to deliver it to that user. The objective is to minimize the maximum time the users have to wait to receive the requested files. Alternatively, we consider maximizing the normalized rate the users are received their requested files. We characterize the optimal normalized rate within a constant gap, independent of the parameters of the problem. First, we consider a deterministic channel equivalent to the original Gaussian problem, and characterize its exact rate-memory trade-off. Then, we rely on intuition developed from the solution of deterministic channel model to develop achievable schemes and converse for Gaussian channel and characterize its rate-memory trade-off within a constant gap (i.e., within 3.58 bits per channel uses). Simulation results show that the gap indeed is less than one bit per channel use. To do this we follow two steps. First we consider an equivalent deterministic channel model and derive exact rate-memory tradeoff. In other words we calculate the rate capacity of deterministic model for two users. Then by using the intuitions gained by solving deterministic model, we try to find the gap of Gaussian model and show this gap is constant. After two users we consider three users and we develop the rate-memory trade-off for deterministic channel. Although the traditional caching does not work for three users deterministic channel, we introduce new caching. The idea of this new caching technique is to first consider two weaker users and implement caching scheme for these two users, then consider the strongest user and update our scheme
  9. Keywords:
  10. Caching ; Multiuser Channel ; Gaussian Broadcast Channel ; Deterministic Channel

 Digital Object List