Page 35 - ITU Journal Future and evolving technologies – Volume 2 (2021), Issue 2

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 2

CLUSTERING‑BASED DEVICE‑TO‑DEVICE DISCOVERY AND CONTENT DELIVERY IN WIRELESS
NETWORKS

1
Tolga Girici and Ahmet Cihat Kazez 1
1 Electrical and Electronics Engineering, TOBB University of Economics and Technology, Ankara, Turkey

NOTE: Corresponding author: Tolga Girici, tgirici@etu.edu.tr

Abstract – Device‑to‑device (D2D) cache placement and content delivery is a recent technology that exploits idle memory
capacity of mobile devices and promises to ful ill users’ content needs with very low delay and without using the scarce cellular
network capacity. Recent literature on D2D caching mostly concentrated on optimal cache placement and mostly disregarded
the required discovery mechanisms and the accompanying overhead. In this work we propose a clustering‑based procedure
that is used to cache content at the helper nodes along with advertisement and discovery procedures for content delivery. We
implement a Long Term Evolution (LTE)‑based simulation setup to evaluate the proposed schemes. Numerical results reveal
that the clustering‑based user and content discovery signi icantly improves cellular of loading and reduces the discovery
overhead with a minimal sacri ice from delay.
Keywords – Advertisement, caching, content delivery, device‑to‑device communication, discovery, of loading

1. INTRODUCTION mission is exempt from RAN and core network traf ic, re‑
quested content must be cached by the edge devices be‑
Recent technological improvements have triggered the forehand and delivered upon request [5].
widespread use of connected devices and caused a dra‑ In the literature there are several studies that investigate
matic increasein mobiledata demand. While meetingthis the cache placement and content delivery for D2D based
demand service providers must also ful ill users’ essential networks with an aim of optimizing throughput, hit rate
Quality of Service (QoS) requirements such as high data and delay with effective usage of limited cache capacity of
rate and low latency. This exigency is conducive to the edge devices [6, 7, 8, 9, 10, 11, 12, 13].
birth of new technologies in next generation communi‑
cation systems, particularly for 5G and beyond. Device‑ Widespread adoption of the D2D technology requires ef‑
to‑device (D2D) communication is one of these technolo‑ icient, practical, and implementable schemes for cache
gies which enables direct communication of nearby users placement and content discovery, along with new busi‑
without any Base Station (BS) assistance and alleviates ness models [4]. In the literature, studies on D2D
load on the Radio Access Network (RAN) and core net‑ cache placement mostly considered optimization‑based
work. Bene its of D2D may be listed as high spectral ef i‑ approaches that disregarded practicality aspects. Opti‑
ciency, reduced latency, extended capacity, and improved mization problems are formulated and solved in order to
power ef iciency [1]. D2D communication may utilize li‑ maximize coverage [6], of loading [7], hit ratio [9, 14, 15,
censed (in‑band) and unlicensed (out‑band) spectrum as 16], minimize delay [8, 11] and minimize cost [10], under
well. In‑bandD2Dcommunicationhasbeende inedinthe constraints such as cache capacity [15], delay and cover‑
3rd Generation Partnership Project (3GPP) Release 12 age [16]. An optimal cache placement problem that aims
and adopted to existing Long Term Evolution (LTE) tech‑ to maximize of loading has been formulated and solved
nology withthe nameLTEDirect. Inthe scope of 3GPP Rel. jointly with the D2D link establishment scheme in [17].
12 network architecture, discovery procedures, synchro‑
nization, authorization, and group operations are stan‑ There are also game‑theoretic schemes that are proposed
dardized [2]. to maximize revenue [12]. An incentive‑based game theo‑
retical approach for the cache placement problem is pre‑
Today, as an add‑on technology LTE Direct is not prevalent sented in [18]. Another group of work utilized stochas‑
and has limited usage in public services such as health tic geometry, where the distribution of nodes is consid‑
care and security applications. On the other hand, for ered as a spatial random process. The authors in [19,
5G, D2D communication is one of the innate technolo‑ 20, 21, 22, 23] used stochastic geometry and found ran‑
gies and leads to several proximity‑based services, such domized caching (in some works along with randomized
as social networking, emergency communication, Inter‑ scheduling [19] ) policies in order to maximize through‑
net of Things (IoT) enhancement, commercial and media put. However, these works only reveal some fundamen‑
services [3]. Cache placement and content delivery are tal trade‑offs between throughput and node density and
considered vital applications of D2D communication [4]. do not give the performance in an actual network, as the
Since D2D technology is device‑centric and data trans‑ actual node distribution may not be a Poisson point pro‑

© International Telecommunication Union, 2021 21