الفهرس 
صفحة العنوانOnly 14 pages are availabe for public view
 |  | from | 177 |  |  |
| | | from | 177 | | |
Abstract
Content caching in Device-to-Device (D2D) wireless cellular network can be
considered an attractive solution to decrease network load during peak time hence
improve network performance. The literature has a strong evidence that user
behavior is highly predictable as well as mobility patterns of users. Cellular
networks can track, learn and predict user preferences and mobility patterns. In such
network, predicting users’ movement pattern allows proactive caching to alleviate
its congestion leading to load decrease. Optimizing caching process is one important
solution to enhance network performance and thus increase offloading probability.
Caching some data items in the off-peak times moves some of the network load to
user nodes and reduces service cost. Moreover, D2D communications permit users
to deliver their pro active dewnloads with requester users in their neighborhood area.
Therefore, users can find their request data either with other helper users around
them or in their local caches. Nonetheless, harnessing user mobility information
enhances networks caching decisions and reduces its cost. The user’s trajectories
information allows the network to predict their presence probability in some popular
locations. Finding an optimal caching strategy eases the network congestion in these
popular locations and enhances the network performance.
This thesis introduces a study to cover the capabilities of D2D caching cellular
networks and investigates how to improve its performance. The research consists of
three main directions, in first, take advantage of user behavior predictability to
offload the network data. Secondly, beneficially the relations between users to
introduce a content storing and sharing among other users. Lastly, leveraging the
information about user locations and hence Quality of Service (QoS) parameters to
enhance caching strategy and the overall throughput. Starting by investigating how
to contribute the user behavior predictability to store some data files during .
times for a possible request throughout peak times. A caching policy strategy is
introduced where the network jointly recognizes group mobility and user
preferences using rewarding system to solve the caching optimization problem.
Hence, optimization algorithms is used to minimize the overall network load by
optimizing the amount of data cached in users devices. Further, the relations
between users’ requests and QoS parameters are introduced in terms of Signal to
Interference plus Noise Ratio (SINR) and energy consumed from helper users’
batteries. Simulations are carried out to evaluate performance of the presented
optimal caching policy with and without QoS constraints. Moreover, joint channel
and power allocation for D2D communications underlaying 5G networks is
discussed including caching policy consideration. Here is, Uplink (UL) resource
reuse between multiple Cellular Users (CUs) and multiple D2Ds connections
located in single Main Base-Station (MBS) and multiple Small Base Stations (SBSs)
is investigated. In order to gain from allocating the UL spectrum, resource and power
allocation algorithm is designed to maximize the overall network sum rate while
guaranteeing QoS requirements for both CUs and D2D links. Finally, simulation
results show that the proposed algorithm efficiently improve the overall network
performance under the consideration of different network parameters.