Page 17 - ITUJournal Future and evolving technologies Volume 2 (2021), Issue 1

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 1

DYNAMIC POWER CONTROL FOR TIME‑CRITICAL NETWORKING WITH HETEROGENEOUS
TRAFFIC

1
2
Emmanouil Fountoulakis , Nikolaos Pappas , Anthony Ephremides 1,2
1 Department of Science and Technology, Linkoping University, Sweden, Electrical and Computer Engineering
2
̈
Department, University of Maryland, College Park, USA
NOTE: Corresponding author: Emmanouil Fountoulakis, emmanouil.fountoulakis@liu.se

Abstract – Future wireless networks will be characterized by heterogeneous traf ic requirements. Examples can be low‑
latency or minimum‑througput requirements. Therefore, the network has to adjust to different needs. Usually, users with
low‑latency requirements have to deliver their demand within a speci ic time frame, i.e., before a deadline, and they coexist
with throughput oriented users. In addition, mobile devices have a limited‑power budget and therefore, a power‑ef icient
scheduling scheme is required by the network. In this work, we cast a stochastic network optimization problem for minimiz‑
ing the packet drop rate while guaranteeing a minimum throughput and taking into account the limited‑power capabilities
of the users. We apply tools from Lyapunov optimization theory in order to provide an algorithm, named Dynamic Power
Control (DPC) algorithm, that solves the formulated problem in real time. It is proved that the DPC algorithm gives a solu‑
tion arbitrarily close to the optimal one. Simulation results show that our algorithm outperforms the baseline Largest‑Debt‑
First (LDF) algorithm for short deadlines and multiple users.

Keywords – Deadline‑constrained traf ic, dynamic algorithms, heterogeneous traf ic, Lyapunov optimization, power‑
ef icient algorithms, scheduling.

1. INTRODUCTION nel conditions. However, many devices may have a lim‑
ited power budget. Therefore, energy‑ef icient commu‑
5G and beyond networks are poised to support a mixed nications have become a very important issue. In this
set of applications that require different types of services. work, we propose a scheduling algorithm that handles
There are two main categories of applications. The irst a heterogeneous set of users with heterogeneous traf‑
category includes applications that require bandwidth‑ ic. In particular, we consider a network with deadline‑
hungry services and the second includes delay‑sensitive constrained users and users with minimum‑throughput
applications. The second category differentiates the cur‑ requirements, with a limited‑power budget. We provide
rent networks from future networks. These applications
an algorithm that solves the scheduling problem in real
require low‑latency services and increase the need for
time. We prove that the obtained solution is arbitrarily
time‑critical networking. In time‑critical networking, ap‑
close to the optimal.
plications are required to deliver their demands within a
speci ic time duration [1]. In other words, each packet
or a batch of packets has a deadline within which data 1.1 Related works
must be transmitted, otherwise, it is dropped and re‑
moved from the system [2]. This is connected with the no‑ Delay‑constrained network optimization and perfor‑
tion of timely throughput. Timely throughput measures mance analysis have been extensively investigated [6]. A
the long‑term time average number of successful deliv‑ variety of approaches have been applied to different sce‑
eries before the deadline expiration [3, 4]. Each time‑ narios. There is a line of work that considers the control
critical application belongs to a different category. For of the maximum number of retransmissions before the
example, motion control, smart grid control, and process deadline expiration. In [7], the authors consider a user
monitoring belong to the industrial control category. Fur‑ transmitting packets over a wireless channel to a receiver.
thermore, the growing popularity of real‑time media ap‑ An optimal scheduling scheme is proposed that provides
plications increases the need for designing networks that the optimal number of retransmissions for a packet. In
can offer services with low latency. Such applications are [8], the authors consider users with packets with dead‑
media production, interactive Virtual Reality (VR), cloud lines in a random‑access network. They show how the
computing, etc, that are under the umbrella of the Tactile number of maximum retransmissions affects the packet
Internet [5]. drop rate. In [9], the authors consider a single transmit‑
With the pervasiveness of mobile communications, such ter that transmits symbols to a receiver. Each symbol
applications need to perform over wireless devices. In or‑ has a deadline and a corresponding distortion function.
der to achieve reliable communication, the devices have The authors consider the distortion‑minimization prob‑
to adapt their power transmission according to chan‑ lem while ful illing deadline constraints. In [10], the au‑

© International Telecommunication Union, 2021 1