A new link activation policy for latency reduction in 5G integrated access and backhaul systems

Full Text

1. Introduction The digitalization of many areas of human activity relies upon a communication system capable of providing a wide range of services. The 5th generation (5G) mobile networks enable the provision of different services including Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communications (URLLC), and Massive Machine-Type Communications (mMTC). The services provided by 5G networks require improvements in various performance indicators. For example, eMBB needs to offer high throughput (up to 10 Gbps) and support high mobility devices (up to 500 km/h). URLLC requires delay reduction down to one millisecond. Finally, for mMTC services, the number of connected devices must be increased to up to 10 million per square kilometer, while also improving their energy efficiency [1]. In order to provision the required performance indicators in 5G, significant changes have been made to the architecture and operations of the 5G core (5GC) and radio access networks (RAN). For example, flexibility and adaptability in synchronization procedures, as well as the allocation and splitting of bands into subcarriers, have been increased. Additionally, modulation, coding, and error correction have been improved [2]. In addition to enhancing the RAN functionality, an important technical innovation of 5G is its substantially expanded frequency range. This allows for higher throughput by allocating vast bandwidth at high frequencies (greater than 24 GHz), while maintaining wide coverage through the utilization of lower frequencies. It is worth noting though that communications in the new highfrequency spectrum suffer from high propagation losses and require significant capital expenditures for upgrading and expanding network hardware infrastructure. In particular, as the coverage area of a base station is reduced due to propagation issues, network densification is necessary, which involves increasing the number of access points (APs) per unit area. © Zhivtsova A. A., Beschastnyy V. A., 2024 This work is licensed under a Creative Commons Attribution 4.0 International License https://creativecommons.org/licenses/by-nc/4.0/legalcode One way to densify 5G networks is to utilize the Integrated Access and Backhaul (IAB) technology. It employs relay nodes that are not wired connected to the core network as additional APs. The interference issues in the resulting multi-hop wireless network call for the half-duplex transmission, meaning that no network node can receive and transmit data at the same time. In turn, a half-duplex system requires an efficient link activation policy, which determines over which links data can be transmitted at any given time. In this paper, we aim to design a new link activation policy for 5G IAB networks that allows for packet delay reduction and throughput maximization and can be employed in both centralized and distributed manners. The rest of the paper is structured as follows. First, in Section 2, we discuss the IAB technology and briefly overview the related work. Then, we formalize the model of an IAB network in Section 3 and propose a new link activation policy in Section 4. Next, in Section 5, we obtain realistic simulation parameters and numerically evaluate performance of the proposed policy in comparison with well-known link activation algorithms. Conclusions are drawn in the last section. 2. Background and related work To minimize capital expenditures in deploying dense 5G networks, the 3GPP (3rd Generation Partnership Project) standardization body has proposed the IAB [3]. IAB allows to use relay nodes that are not directly connected to the core network as relaying APs. As depicted in Figure 1, there are two types of APs in an IAB network: an IAB donor directly connected to the core network by a wired link, and one or more IAB nodes which transmit traffic from or to the core network through the IAB donor. The wireless links in the IAB network are divided into two types: access links between an AP and a User Equipment (UE), and backhaul links between APs. Both types of links use a shared time-frequency resource, as the name of the technology implies. The IAB technology is based on the distributed architecture of 5G networks. This architecture separates the layers of the data transfer protocol stack between central and distributed units, as shown in Figure 1. A Distributed Unit (DU) implements Radio Link Control (RLC), Medium Access Control (MAC), and Physical Layer (PHY). The DU is present at each AP and ensures the establishment, maintenance, and termination of radio connections. The Central Unit (CU) implements Service Data Adaptation Protocol (SDP) and Packet Data Convergence Protocol (PDCP). The CU is only present in the donor and provides connection with the core network. Each IAB node contains a Mobile Termination (MT). This component supports the Backhaul Adaptation Protocol (BAP), which forwards data streams that travel through multiple IAB nodes to and from the IAB donor. In the first IAB standardization document [3] released in 2018, the IAB network was defined as a multi-hop wireless network with static APs and the ability of path selection. Also, the standard provides a list of possible options for implementation. For example, either in-band or out-of-band backhauling can be used. The use of time, frequency, or spatial multiplexing is permitted, as is endto-end or hop-by-hop automatic repeat request (ARQ). The resource allocation is not fully determined by the standard, and has been explored in various research projects. For an extensive review, see [4]. As previously mentioned, the IAB standard allows the simultaneous operation of access and backhaul links within the same frequency band. This reduces the downtime for the radio resources, but also increases interference [5]. Each transmitter interferes with all other active receivers in the network, except the one it is communicating with. The high-frequency 5G spectrum allows for directional transmission, reducing interference in many channels. Nevertheless, interference that occurs during simultaneous reception and transmission remains significant [6]. To eliminate interference caused by simultaneous reception and transmission in the IAB network, the standard [3] recommends using the half-duplex mode. This mode helps to reduce interference by limiting the number of channels on which transmission occurs at any given time. More precisely, halfduplex mode prevents any AP in the IAB network from receiving and transmitting data simultaneously. Although the half-duplex mode limits the network throughput and increases delays, it is an effective and simple way to reduce interference. To efficiently implement half-duplex, it is essential to schedule transmission over links. This can be done by dividing time into slots and marking each link with 1 (ON) if it is allowed to transmit in the slot and 0 (OFF) otherwise [7-10]. Such link scheduling permits to ensure that the half-duplex constraints are met and to optimize selected performance metrics. For example, in [9] the link scheduling algorithm maximizes minimal user throughput, in [10] it optimizes the sum of user throughputs, and in [7, 8, 11] it targets some convex function of user throughput (such as the sum of logarithms). In [7-11] the link scheduling is performed by solving an optimization problem with the objective function of throughput. On the other hand, constructing a queuing model of the studied network allows to evaluate and optimize the delay [12, 13], as well as to prove the stability of the network under some scheduling algorithms with any acceptable rates of incoming traffic [11, 14]. This approach was used to derive a number of link scheduling algorithms for general multi-hop wireless networks with interference, and in particular several throughput optimal greedy dynamic algorithms for efficient centralized control of multi-hop networks, which choose a transmission mode based on the current system state via argmin or argmax. Backpressure [15] is the most recognized throughput-oriented algorithm for network control and can be utilized for link scheduling, routing or flow control problems [11, 16-18]. While backpressure handles queue lengths, such algorithms as the largest weighted delay first [19, 20], oldest cell first [21] and delay-based backpressure [22] use packet delays to specify the system state. The latter is the delay-based version of backpressure and allows to reduce the maximum packet delay in the original backpressure algorithm. The

About the authors

Anna A. Zhivtsova

RUDN University

Email: aazhivtsova@sci.pfu.edu.ru
ORCID iD: 0009-0007-8438-6850

bachelor’s degree student of Department of Probability Theory and Cyber Security

6 Miklukho-Maklaya St, Moscow, 117198, Russian Federation

Vitaly A. Beschastnyy

RUDN University

Author for correspondence.
Email: vbeschastny@sci.pfu.edu.ru
ORCID iD: 0000-0003-1373-4014
Scopus Author ID: 57192573001

Candidate of Physical and Mathematical Sciences, assistant professor of Department of Probability Theory and Cyber Security

6 Miklukho-Maklaya St, Moscow, 117198, Russian Federation

References