Two stage downlink scheduling for balancing QoS in multihop IAB networks
Introduction
Ultra-dense deployment combined with millimeter-wave (mmWave) communication has emerged as an effective solution to realize the vision of Fifth Generation (5G) mobile networks [1], [2]. The mmWave spectrum offers enormous bandwidth to achieve high data rate and low delay, which are essential for supporting diverse services. Although mmWave communication introduces severe path and penetration losses, it also supports a massive antenna array to achieve high directivity links, and beamforming [1]. An ultra-dense deployment achieves better frequency reuse, energy efficiency, and more line-of-sight links due to reduced inter-site distance. Having said that, ultra-dense deployment may not be an economical choice for operators as they need to connect each cell to the 5G Core Network (5GC) using a wired backhaul (e.g., optical fiber or digital subscriber line). The deployment of wired backhaul requires not only considerable expenditure but also significant installation time. Moreover, wired backhaul installation might not be possible/allowed in certain areas.
Integrated Access and Backhaul (IAB) has been envisaged as a scalable and cost-effective solution to overcome geographical constraints in ultra-dense deployment [3], [4], [5]. In IAB, a Base Station (BS) shares the same spectrum to serve User Equipment (UE) in access links and to communicate with other BSs in wireless backhaul links. Sharing the spectrum or radio resources can be done in either time, frequency, or spatial domain. In the IAB network, wired backhaul to the 5GC is only available at a few specific BSs, and these BSs act as access gateways for other BSs having wireless backhauls. Thus IAB enables a network to form a multihop backhaul topology where a UE can communicate with the 5GC over any number of wireless backhaul links. An ultra-dense network typically produces massive backhaul traffic in the 5GC, which means a wireless backhaul requires high bandwidth and reliability similar to those of a fiber connection. This issue is attempted to be addressed in IAB by utilizing mmWave spectrum for backhaul as well as access. Therefore, IAB reduces reliance on wired backhaul availability at each BS, enabling an operator to provide a faster and more flexible rollout of mmWave 5G networks with significantly reduced deployment cost.
Keeping in mind these merits, the 3rd Generation Partnership Project (3GPP) has recently standardized an IAB architecture in Release 16 [6]. Consequently, both operators and users expect 5G services to be available with the IAB network. It means IAB network should meet 5G specifications that are defined for a single hop 5G network. Nonetheless, the IAB feature poses many open research challenges that are still not addressed. In particular, radio resource scheduling is widely believed to be a valuable tool for using available spectrum and improving overall system throughput efficiently in a wireless network. In literature, many channel-aware radio resource schedulers have been proposed. The gains of these channel-aware schedulers, however, may be limited in IAB networks due to several reasons, namely IAB network topology, multihop distances between UEs and the 5GC, and sharing of spectrum by access and backhaul links. Specifically, multihop relaying makes resource scheduling more challenging as access traffic passes through multiple wireless backhaul links, impacting network and UE performance. In fact, UEs that are far away from IAB-donors may not have sufficient throughput for basic services. The issue worsens if UEs need services with different Quality of Service (QoS) requirements, and the network is expected to guarantee each of them. Thus, the resource scheduler must be aware of IAB topology, QoS requirements, and backhaul constraints for its practical implementation and improve network performance. This begets the need for control information exchange among IAB-nodes. Designing a scalable control exchange mechanism is also a challenge.
Motivated by the aforementioned scheduling issues, we propose a two-stage QoS-based and channel-aware downlink scheduler tailored for IAB networks. The scheduler aims to fulfill QoS requirements for different service types and to achieve fairness across UEs belonging to the same service type. We describe both stages of the scheduler in detail with its underlying principle for QoS provisioning. The first stage schedules access and backhaul links at a BS. The second stage distributes traffic volume allocated for a backhaul link to its underlying downstream links. Because these two stages have different provisions, we propose different algorithms for each stage. Furthermore, there is no consensus on how we redefine or adapt bearer (or QoS) requirements to a multihop IAB network. We have addressed this issue by redefining minimum bitrate and delay constraints at each hop for every downstream UE after carefully considering factors like hopcount, service type, and status of other UEs. In the end, the BS scheduler considers the QoS requirements of every downstream (direct or indirect) UEs irrespective of their topological location within the network. We design our scheduler for many-to-one flow mapping. However, as an extension, we further provide a methodology to adapt the scheduler for 1:1 mapping scenario in case the operator prefers to use such mapping. The main contributions of this paper are summarized as follows:
- •
Design scheduling frameworks for different services based on their QoS requirements and service types. For example, a service requiring minimum bitrate and delay constraints has a different scheduling framework than a service with either or none of these constraints.
- •
Devise delay frameworks for different types of services. These frameworks are probabilistic and follow the delay definition according to the 3GPP specification.
- •
Propose a simple rate adaptation technique that periodically adapts sending rate based on minimum bitrate and congestion level in the network to improve resource efficiency and control any unwarranted increase in delays.
Section snippets
Related works
In recent times, both academia and industry researchers have shown major interest in IAB. These research activities have demonstrated the feasibility, potentials and challenges of mmWave-based IAB networks using either end-to-end simulations [7], field trials [4], [8] or optimal association and routing [5], [9], [10]. Specifically, on the resource scheduling problem, many existing solutions are based on network throughput maximization [2], [11], [12], [13], [14]. Others have formulated resource
An overview of 3GPP IAB
An IAB network has two Radio Access Network (RAN) elements, namely IAB-donor and IAB-node. An IAB-donor is similar to a next-generation Node B (gNB) and is connected to the 5GC using a wired connection. In contrast, an IAB-node can only wirelessly backhauls UE or access traffic, possibly through multiple hops to IAB-donor. The IAB-donor then forwards access traffic to the 5GC. In the downlink, the IAB-donor first backhauls access traffic to an appropriate child IAB-node. Then IAB-node either
5G QoS model
The 3GPP 5G QoS model supports a diverse range of services with different QoS requirements [29]. A QoS flow is primarily defined with nine attributes – 5G QoS Identifier (5QI), Priority Level, Resource Type, Guaranteed Flow Bit Rate (GFBR), Maximum Flow Bit Rate (MFBR), Maximum Data Burst Volume (MDBV), Averaging Time Window (ATW), Packet Delay Budget (PDB) and Packet Error Rate (PER).
The 5QI determines the packet forwarding treatment (such as queue management and RLC layer configuration) that
Bearer mapping in IAB
In an IAB network, a BS needs to multiplex UE flows onto backhaul RLC channels, and generally speaking, two options on flow mapping are possible [3]. In one-to-one (or 1:1) mapping, each UE flow is mapped to a separate backhaul RLC channel. Many-to-one (or :1) mapping, on the other hand, multiplexes several flows with similar QoS profiles onto a single backhaul RLC channel, even if they belong to different UEs. Consequently, each backhaul RLC channel carries a different number of UE flows. In
Resource scheduling challenges in IAB
One of the prime challenges of the IAB resource scheduler is to guarantee GFBRs and maintain fairness across different QoS flows irrespective of their topological locations within a network. Enforcing these aspects is challenging at IAB-node for :1 mapping because its parent node aggregates traffic for multiple UEs into a single backhaul flow before forwarding it. As mentioned earlier, BSs can obtain information about downstream QoS flows carried on each backhaul flow. The BSs may additionally
System model
There may be multiple IAB-donors in a deployment or geographic region, each acting as an access gateway(or root node) for their respective IAB topologies. These independent IAB network topologies, however, behave similarly from a resource management perspective. It means that for analysis and understanding purposes, it is adequate to focus on an IAB topology comprising a single IAB-donor and multiple IAB-nodes that are located at different hop levels from the 5GC. We describe our downlink
QoS based scheduling framework for each resource type
This section presents scheduling framework for different resource types according to their 5QI characteristics and 3GPP specifications. The objective of such scheduling framework is to achieve the following:
- 1.
Satisfy GFBRs of GBR or DGBR flows.
- 2.
Satisfy PDB of each QoS flow.
- 3.
Maximize throughputs of NGBR flows.
We accomplish the objective by balancing QoS in such a way as to avoid starving NGBR flows while prioritizing GBR and DGBR flows, especially in the non-saturated state. We further strive for at
Proposed two stage QoS flow based downlink scheduler for IAB
This section provides details of the proposed two stage QoS-based downlink scheduler. The scheduler combines scheduling frameworks for different resource types described in the last section. The scheduling algorithm comprises two stages; the operation of each stage is outlined below.
Adaptation of :1 mapping for 1:1 mapping
Since each downstream QoS flow has a separate backhaul flow in 1:1 mapping, we can suppose that these flows are directly connected to the BS. It means that we do not need to perform stage II scheduling. However, we do need to adapt stage I with minor tweaks to make it suited for 1:1 mapping in a multihop setting. After taking hop distances into account, the adapted utilities for WPF and PD algorithms are given as The rest of the
Feasible implementation of scheduler
The proposed scheduler needs the knowledge of several pieces of information available to BSs for its proper working. Since the IAB standard is still under development, we have identified possible ways to collect and maintain these scheduling information and presented them in Table 1. The ‘Details’ column provides how the specified information is being made available to the relevant BSs through existing 3GPP protocols. The table also provides details on how the information can be made available
Performance evaluation
We consider urban micro deployment [3] with hexagonal cell sites for one IAB-donor and IAB-nodes. There are UEs dropped uniformly and randomly within each cell. The BSs are equipped with uniform rectangular antenna arrays with elements on both transmitter and receiver sides. In contrast, each UE has only one element. For physical layer aspects of mmWave frequencies, we use NYU model [1], [34] for evaluating beamforming gains. Half power beamwidth in both azimuth and elevation
Concluding remarks
Many researchers believe that deploying ultra-dense networks is the primary way to realize capacity goals envisaged for the forthcoming 5G technology. However, connecting every cell to 5GC through a wired backhaul is not economical for operators. In this regard, IAB may prove to be a promising solution to enable faster and economical deployment of ultra-dense 5G cells. IAB provides wireless backhaul links to BSs and facilitates them to relay access traffic across the network. Further, IAB also
CRediT authorship contribution statement
Shashi Ranjan: Conceptualization, Validation, Writing, Software. Pranav Jha: Conceptualization, Validation, Writing – review & editing. Abhay Karandikar: Conceptualization, Writing – review & editing, Supervision, Project administration, Funding acquisition, Resources. Prasanna Chaporkar: Conceptualization, Writing – review & editing, Supervision.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
This work has been supported by the Ministry of Electronics & Information Technology (MeitY), India as part of the project ‘Next Generation Wireless Research and Standardization on 5G and Beyond’.
Shashi Ranjan is currently pursuing Ph.D. from IIT Bombay, India. He completed his MTech from IIT Bombay in 2013 and BTech from BIT Mesra, India in 2010. During the 2013–2015 period, he worked as Research Engineer at IIT Bombay on the project Dual Connectivity LTE-WiFi Solution for Broadband Wireless funded by MeitY. He has been an IIT Bombay representative at 3GPP RAN2 and SA6 Working Group. His research interests include wireless backhaul, network architecture, radio resource management and
References (36)
- et al.
Co-optimizing performance and fairness using weighted PF scheduling and IAB-aware flow control
IEEE Wirel. Commun. Netw. Conf. (WCNC)
(2020) - et al.
Service-differentiated downlink flow scheduling to support QoS in long term evolution
Comput. Netw.
(2016) - et al.
Analysis of the increase and decrease algorithms for congestion avoidance in computer networks
Comput. Netw. ISDN Syst.
(1989) - et al.
Millimeter wave channel modeling and cellular capacity evaluation
IEEE J. Sel. Areas Commun.
(2014) - et al.
Millimeter wave integrated access and backhaul in 5G: Performance analysis and design insights
IEEE J. Sel. Areas Commun.
(2019) Study on integrated access and backhaul
(2019)- et al.
Integrated access and backhaul: A key enabler for 5G millimeter-wave deployments
IEEE Commun. Mag.
(2021) - et al.
On integrated access and backhaul networks: Current status and potentials
IEEE Open J. Commun. Soc.
(2020) NR and NG-RAN overall description
(2021)- et al.
Integrated access and backhaul in 5G mmwave networks: Potential and challenges
IEEE Commun. Mag.
(2020)
Field trial on millimeter wave intergrated access and backhaul
IEEE Vehic. Technol. Conf. (VTC-Spring)
Integrated access backhaul in millimeter wave networks
IEEE Wirel. Commun. Netw. Conf. (WCNC)
Distributed path selection strategiesfor integrated access and backhaul at mmwaves
IEEE Glob. Commun. Conf. (GLOBECOM)
Bandwidth partitioning and downlink analysis in millimeter wave integrated access and backhaul for 5G
IEEE Trans. Wireless Commun.
Optimal link scheduling in millimeter wave multi-hop networks with MU-MIMO radios
IEEE Trans. Wireless Commun.
Deep reinforcement learning-based spectrum allocation in integrated access and backhaul networks
IEEE Trans. Cogn. Commun. Netw.
RL-based resource allocation in mmwave 5G IAB networks
Mediterranean Commun. Comput. Netw. Conf. (MedComNet)
Discrete power control and transmission duration allocation for self-backhauling dense mmwave cellular networks
IEEE Trans. Commun.
Cited by (0)
Shashi Ranjan is currently pursuing Ph.D. from IIT Bombay, India. He completed his MTech from IIT Bombay in 2013 and BTech from BIT Mesra, India in 2010. During the 2013–2015 period, he worked as Research Engineer at IIT Bombay on the project Dual Connectivity LTE-WiFi Solution for Broadband Wireless funded by MeitY. He has been an IIT Bombay representative at 3GPP RAN2 and SA6 Working Group. His research interests include wireless backhaul, network architecture, radio resource management and rural broadband.
Pranav Jha has extensive research and development experience in communication and networking technologies with expertise in network architecture, communication protocols and resource management algorithms. He currently works with IIT Bombay, Mumbai, India where he focuses on research programs under the broad umbrella of wireless networking with emphasis on SDN based architectures for Mobile Networks, Public Safety Communication and Rural Broadband Communication. He has multiple research papers, patents issued and pending, and contribution to IEEE and 3GPP standards.
Abhay Karandikar is currently the Director of IIT Kanpur (on leave from IIT Bombay). He served as Member (part-time) of Telecom Regulatory Authority of India (TRAI) from 2018 to 2020. In IIT Bombay, he served as Institute Chair Professor in the Electrical Engineering Department, the Dean (Faculty Affairs) from 2017 to 2018 and the Head of the Electrical Engineering Department from 2012 to 2015. Prof. Karandikar is the founding member of Telecom Standards Development Society, India (TSDSI), India’s standards body for telecom. He was the Chairman of TSDSI from 2016 to 2018. His research interests include resource allocation in wireless networks, software defined networking, frugal 5G and rural broadband. Detailed biography can be found at https://www.ee.iitb.ac.in/~karandi/.
Prasanna Chaporkar received the MS degree from the Faculty of Engineering, Indian Institute of Science, Bangalore, India, in 2000, and the Ph.D. degree from the University of Pennsylvania, Philadelphia, Pennsylvania, in 2006. He was a ERCIM postdoctoral fellow with ENS, Paris, France, and NTNU, Trondheim, Norway. Currently, he is a professor in the Indian Institute of Technology, Mumbai. His research interests include resource allocation, stochastic control, queueing theory, and distributed systems and algorithms.