Delay Performance Optimization in Passive Optical Network

(1)

Delay Performance Optimization in Passive Optical Network

Fahmida Rawshan¹, Monir Hossen², and Md. Rafiqul Islam¹

1Dept. of Electrical and Electronic Engineering, Khulna University of Engineering & Technology, Khulna, Bangladesh

2Dept. of Electronics and Communication Engineering, Khulna University of Engineering & Technology, Khulna, Bangladesh Email: [email protected], [email protected], [email protected]

Abstract‒ Delay is one of the most important performance indication parameters for fifth-generation (5G) and beyond communication systems. Passive optical network (PON) is gaining attention and being considered as a conjugating part of optical and wireless technology. This conjugation aims to maintain strict delay requirements in PON systems. This work investigates the relation among various delay performance parameters such as the number of optical network units, number of time slots, throughput, and bandwidth in a PON system. Different delay parameters are analyzed mathematically as well as numerically. The analyses show the ways to optimize the delay parameters. The performance analyses prove that the delay affecting parameters can be optimized as the delay requirement.

Index Terms— 5G, C-RAN, delay, passive optical network, optical network unit, and optimization.

I. Introduction

The exponential growth of various high-data-rate evolving applications, such as artificial intelligence supported services, virtual reality, and the internet of everything, has led to an enormous size of traffic in the fifth-generation (5G) and beyond communication systems [1], [2]. The global mobile traffic size was 7.462 EB/month in 2010, and this is predicted to be 5016 EB/month in 2030 [3]. This scenario certainly clarifies the implication of the enrichment of communication network systems. A passive optical network (PON) will be the crucial part to support this huge volume of data traffic in 5G and beyond (5GB) communication systems [4], [5]. The tremendous increase of different applications such as emails, data, file sharing, messaging, cloud services, video calls, online gaming, and online movie streaming with different levels of quality of service (QoS) requirements, opens the scope to further develop the data transmission systems of PON. The internet services must be able to offer a satisfactory level of delay requirement along with additional bandwidth to satisfy end-user demands.

In 5GB communication systems, the end-to-end (E2E) delay is a very important parameter to satisfy the user’s QoS requirement. The delay requirement of 5G and 6G networks are 10 ms and 1 ms, respectively [2]. However, the delay requirements for different applications are not the same. In some cases, delay requirements are very strict. High-speed communication systems like centralized radio access network (C-RAN) needs high speed and low latency optical

infrastructure to support the fronthaul section. Hence, the PON systems must have the ability to provide a wide range of strict delay requirements.

The delay in a PON system depends on various factors.

Fig. 1 illustrates different factors those influence the total delay in a PON system. Among these, some factors are also considered as performance indicators such as throughput, number of optical network units (ONUs), and number of channels [6]. An increased number of ONUs or channels upsurges the capacity. However, the increase of any of these numbers also increases the delay in the PON system. The increase in throughput through a multi-input-multi-output (MIMO) system also increases the delay in the PON system.

Therefore, it is possible to compromise delay requirements with these factors. Hence, the optimization of these parameters is essential to support a huge volume of data traffic along with strict delay requirements to maximize the resources.

This work provides the relation among various delay performance parameters such as the number of ONUs, number of time slots, throughput, and bandwidth. Different delay parameters are discussed in detail. We analyze different delay parameters mathematically and numerically considering C-RAN. This work shows how to optimize among different delay related parameters and how it affects the optimization.

Delay Throughput

No. of ONUs

Fiber length

Processing time

Slot size Propagation

time No. of

channels

No. of time slots

Fig. 1. Delay and related parameters in a PON system.

551

Authorized licensed use limited to: Khulna Univ of Engineering & Technology - KUET. Downloaded on February 06,2022 at 14:17:19 UTC from IEEE Xplore. Restrictions apply.

(2)

CU

DU DU

OLT OLT OLT

ONU ONU RU RU

Central unit Optical midhaul

Optical fronthaul

Optical fiber Splitter

DU

Fig. 2. A typical PON based C-RAN system.

The rest of the paper is organized as follows. Section II presents a simple PON system architecture. The delay formulation in the PON system is presented in Section III.

Section IV presents the performance analysis of the optimization. Finally, we deliver our conclusions in Section V.

II. PON System Architecture

The C-RAN architecture is an efficient cellular network to support extensive mobile traffic and a candidate to be widely deployed for 5GB networks. Here, base station (BS) functionality is consists through three parts: a central unit (CU), a distribution unit (DU), and a radio unit (RU). The fronthaul section consists of DU and RU. The essential requirement here is to transport digitized air interface signals in fronthaul over an extremely low latency and high bandwidth transport layer. Time and wavelength division multiplexing (TWDM) PON, also known as next-generation PON (NG-PON2) can meet this demand with proper wavelength and bandwidth allocation schemes [7]. It can support fronthaul traffic as well as small cell networks traffic through optical fibers connected to intended cells [8].

A scenario of PON supported fronthaul is provided in Fig.

2. The OLT is located at the central office (CO). A single OLT resides near each DU. While ONUs are placed on antenna sites maintaining connection to a RU. It is shown that the fronthaul network between DU and RU is connected via a PON. In the midhaul section CU manages challenges on synchronization and control aspects placed in the central location and the air interface component. Backhaul refers to the transport of all mobile traffic over IP tunnels.

III. Delay Formulation

The fronthaul system should finish forwarding data within a predefined threshold time. This is defined as a delay threshold. The allowable delay in the PON system (T_max^c −T_start^c ) can constitute considering propagation delay Pn and packet processing delay in ONU, tonu. The delays of

Threshold time

RU1

ONU1

OLT

DU ONU2 RU2

RUN-1

……….……….

tDU pn tonu tRU

ONUN-1

RUN

ONUN Optical fiber

Splitter

Fig. 3. Delay in PON based fronthaul.

the RU to ONU and the OLT to DU are dRU and dDU, respectively, [8]. The upstream link bandwidth is L per channel while considering the PON overhead.

In the upstream case, burst size is determined considering the worst circumstance of transmission. Where the ONUs each can transmit STBNTBNMIMO sizeburst. The size of block for transportation is STB. Each sub-frame contains NTB blocks, and NMIMO signifies MIMO layer number.

Real networks have non-symmetric the delay for the two transmission directions. This case is due to different fiber lengths or the use of different wavelengths in the optical transmission system. However, the source of asymmetry results in an error in the time synchronization distribution. It must be known and compensated for or very small to be negligible. Table 1 shows the categories of time division duplex (TDD) time synchronization requirements.

In Fig. 4, the allowed delay in a PON system is defined by (T_maxc −Tstart^c ). The starting of transmission is denoted by

startc

T . σ is the synchronization timing error. The data can appear in an ONU lately or early (by the value of σ) from desired starting, T_start^c . Thereby, twice the value of σ reduction (from the allowed delay margin) is considered to achieve the collision-free transmission. qn and pn are the queuing and the propagation delay, respectively, for n’th ONU. These two delays can be extracted from the threshold value to calculate the allowable delay. tONU is the processing delay and δ is the time to transmit maximum burst.

Table 1 TDD Time-synchronization requirements [8]

Category Time error value

A+ <12.5 ns

A <45 ns

B <110 ns

C <1.5 µs

552

(3)

3 7

n 2

N 3

1 ... s ... S

t

t λ 1

λ 2

λ 3 startc

T T_max^c −σ

10 1 t

λ 4

δ

T

₁c

s=slot ID n=ONU ID

1 ... s ... S

Fig. 4. A scenario of time slot and ONU distribution per wavelength channel of a PON based fronthaul.

The total delay of n^th ONU, tn can be expressed as:

σ δ ≤ _max− −2 +

+ +

= _onu _n _n ^c _start^c

n t q p T T

t (1)

Let Ts^cdenotes the starting time of s^thslot of cycle number c. It can be expressed as:

δ σ+ + _min+( −1) +

=T d p s

T_s^c _start^c _onu (2)

where pmin denotes the minimum delay for propagation which is shown in Fig. 3. It occurs during data transmission between OLT and ONU2.

S denotes the total time slots per wavelength channel. It is formulated by:

δ σ ^c

c T

S T^max− − ¹

= (3) The maximum slot size is equal to maximum burst transmission time and it can be expressed as:

L N N S_TB _TB _MIMO

δ= (4) Total wavelength channels are proportional to ONU numbers of and it can be expressed as:

S

K=N (5)

1 2 3 4

0 4 8 12 16 20 24

Threshold delay 500 ^µs Threshold delay 250 ^µs Threshold delay 100 ^µs

N_MIMO

Number of time slots

Fig. 5. Number of time slot variation with burst size.

IV. Performance Analysis

We consider specific 5G class transmission parameters which are fit for current PON. We also consider MAC-PHY split which is the 6^th splitting function [9]. The delay parameters for the current analysis are considered as follows:

• RU to ONU delay, tRU= 5 μs

• DU to OLT delay, tDU=15 μs.

• the propagation delay pn for n^th ONU is 5 μs/km

• NG-PON2 [7] considered with four wavelength channels

• upstream bandwidth L is 8.7 Gb/s per channel (overhead considered)

• STB = 97896 bits and NTB = 2

• TDD time-synchronization requirements was set as table 1, error value 1.5μs

Fig. 5. shows the number of time slot investigation.

Maximum burst size δ is changed by changing the MIMO layer. NMIMO was set to 1, 2, 3, and 4. While wavelength number is fixed at K=4 for NG-PON2. The maximum number of time slots is studied for various delay thresholds (i.e., 100 μs, 250 μs, and 500 μs). It is shown that an increase in MIMO layer increases the burst size hence, the number of time slots decreases for the same threshold time per wavelength channel. It is easily understood that the higher delay threshold occurs at the higher number of time slots for lower burst size.

The number of ONUs for delay optimization is investigated in Fig. 6. The maximum burst size is changed by changing MIMO layers. The wavelength number is fixed at K=4 for NG-PON2. The maximum ONU number that can be accommodated in the system is studied for various delay threshold (100 μs, 250 μs and 500 μs).

553

(4)

1 2 3 4 0

14 28 42 56 70 84

Threshold delay 500 ^µs Threshold delay 250 ^µs Threshold delay 100 ^µs

N_MIMO

Maximum number of ONUs

Fig. 6. Limit of ONU number with burst size.

1 2 3 4

0 15 30 45 60 75 90

N_MIMO

Time slots [µs]

Fig. 7. Time slot size variation with burst size.

NMIMO was set to 1, 2, 3 and 4. Increasing NMIMO increases the burst size as a result number of slot per channel decrease with respect to a single threshold. As a consequence, the total number of time slots for K channels is lower and system accommodates lower number of ONUs. It can be depicted that higher the delay threshold it raises the number of time slots as a result maximizes the number of ONUs for lower MIMO layers.

The time slot optimization with the throughput is performed in Fig. 7. NMIMO was set to 1, 2, 3 and 4. The delay threshold was set at 500 μs. As NMIMO increases the burst size gets larger. As a result, the number of timeslots decreases per wavelength channel. While as the timeslots increase in length to achieve a fixed threshold delay.

The DU-RU allowable delay is analyzed in Fig. 8. For different burst sizes, the maximum allowable delay from DU to RU has been investigated with respect to the number of ONUs. It is shown that the higher the burst size, it is to allow the higher delay for the same number of ONUs.

20 40 60 80 100

0 300 600 900 1200 1500 1800 2100 2400

δ=22.5 µs δ=45 µs δ=90 µs

Number of ONUs

Allowable delay [µs]

Fig. 8. DU-RU allowable delay investigation with the number of ONUs. V. Conclusions

PON is one of the most important elements to support 5GB optical-wireless communication systems. The applications in these systems require very strict E2E delay. To support the strict delay requirement, we may need to optimize the performance of other parameters. This work investigates the relationships among various delay performance parameters such as the number of ONUs and time slots, throughput, and bandwidth in a PON supported C-RAN. Mathematical relations among different delay parameters are expressed.

The variations of different delay parameters are also analyzed graphically. The analyses show how the delay parameters can be optimized. This work will surely help to design an efficient PON based fronthaul for the C-RAN system by making optimization among different parameters to support strict delay requirements as well as to maximize the system resources.

Acknowledgement: The fellowship of Fahmida Rawshan was supported by University Grants Communication of Bangladesh through UGC Fellowship for PhD Program.

References

[1] I. F. Akyildiz, A. Kak, and S. Nie, "6G and beyond: The future of wireless communications systems," IEEE Access, vol. 8, pp. 133995- 134030, 2020.

[2] M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y. M. Jang, "6G wireless communication systems: Applications, requirements, technologies, challenges, and research directions," IEEE Open Journal of the Communications Society, vol. 1, pp. 957 - 975, Jul. 2020.

[3] ITU-R M.2370-0, IMT traffic estimates for the years 2020 to 2030, Jul. 2015.

[4] H. S. Abbas and M. A. Gregory, "The next generation of passive optical networks: A review," Journal of Network and Computer Applications, vol. 65, pp. 53-74, May 2016.

[5] R. Kaur et. al, "Resource allocation and QoS guarantees for real world IP traffic in integrated XG-PON and IEEE802.11e EDCA networks," IEEE Access, vol. 8, pp. 124883-1248, Jul. 2020.

[6] Glen Kramer, Ethernet Passive Optical Networks, New York:

McGraw-Hill, 2005.

[7] D. Nesset, "NG-PON2 technology and standards," Journal of Lightwave Technology, vol. 33, no. 5, pp. 1136-1143, Mar. 2015.

[8] Y. Nakayama and D. Hisano, "Wavelength and bandwidth allocation for mobile fronthaul in TWDM-PON," IEEE Transactions on Communications, vol. 67, no. 11, pp. 7642-765, Nov. 2019.

[9] E-UTRA Physical Layer Procedures, document 3GPP TR 36.213 Release 14.4.0, Sep. 2017".

554