Queue-Sensing-Based CSAT: A Downlink Transmission

Scheme for LTE-M in Unlicensed Spectrum

Zelin Gong

, Ying Wang

and Tao Luo

Beijing Key Laboratory of Network System Architecture and Convergence,

Beijing University of Posts and Telecommunications, Beijing, P.R. China

Beijing Laboratory of Advanced Information Networks,

Beijing University of Posts and Telecommunications, Beijing, P.R. China

Keywords: LTE-M, LTE-U, Duty cycle, Unlicensed spectrum.

Abstract: A shortage of licensed spectrum is a major hindrance to the capacity of Long Term Evolution for Metro

(LTE-M) to carry urban rail transit services at present. Inspired by LTE-Unlicensed (LTE-U) offloading

some of data traffic transmitted in licensed spectrum by accessing the unlicensed 5GHz frequency band, we

propose a queue-sensing-based Carrier Sensing Adaptive Transmission (CSAT) scheme for LTE-M in

unlicensed spectrum based on LTE-U. The scheme calculates the minimum amount of data that should be

transmitted during current transmission duration by queue length at the beginning of each LTE-U cycle,

according to which, LTE-U adjusts its duty cycle to meet the requirement of LTE-M service. This paper

focuses on the downlink transmission scheme for LTE-M in unlicensed spectrum, and verifies its

performance by simulation, which tells that Queue-sensing-based CSAT will not only provide a reliable

transmission for LTE-M services, but also maximize the fairness of channel occupancy in unlicensed

spectrum.

1 INTRODUCTION

LTE-M is a new generation of train-ground commu-

nication system for urban rail transit integrated

service based on LTE. According to the allocation of

the wireless frequency band for urban rail transit in

Beijing, LTE-M is allowed to use 1785-1795MHz

band with the bandwidth of 10MHz in the

aboveground area, while 1785-1805MHz band is

allowed to use in the underground area (Beijing

Radio Administration, 2017). Additionally, LTE-M

specification requires a structure of dual-network

redundancy to ensure reliable transmission of

Communication Based Train Control (CBTC)

service, where the A network carries multiple types

of service, including CBTC, Passenger Information

System (PIS) and Closed-circuit Television (CCTV),

and the B network only carries CBTC service

serving as a backup network (China Urban Rail

Transit Association, 2016). Through the test of

Hangzhou subway line 4, it can be seen that LTE-M

integrated service system requires a bandwidth at

least 10MHz (Shao and Xie, 2017), and the available

spectrum resource in the aboveground area of

Beijing cannot meet the requirements of LTE-M for

dual-network redundancy and integrated service

bearer. To solve this problem, a solution that

offloading some of LTE-M service transmitted in

licensed spectrum onto the unlicensed 5GHz fre-

quency band has been proposed.

Many literatures focus on the coexistence of

LTE and Wireless Local Area Network (WLAN) in

unlicensed spectrum. For example, LTE based on

Listen Before Talk (LBT) avoids channel collision

through a WLAN-like competitive channel access

mechanism (Hu et al, 2016). The duty cycle solution

enables LTE and WLAN to access the channel in

turn by time division multiplexing technology (Choi

and Park, 2015). CSAT improves the duty cycle

solution to enable LTE to dynamically adjust the

duty cycle of LTE according to the current channel

load (Qualcomm, 2014). At present, the two main

technology for LTE in unlicensed spectrum are

LTE-U based on the duty cycle mechanism and

Licensed Assisted Access (LAA) based on LBT

mechanism. Compared with the competitive channel

access mechanism of LAA, the duty cycle

mechanism of LTE-U is more convenient to control

Gong, Z., Wang, Y. and Luo, T.

Queue-Sensing-Based CSAT: A Downlink Transmission Scheme for LTE-M in Unlicensed Spectrum.

DOI: 10.5220/0008098500920097

In Proceedings of the International Conference on Advances in Computer Technology, Information Science and Communications (CTISC 2019), pages 92-97

ISBN: 978-989-758-357-5

the transmission performance of LTE-M service by

adjusting the duty cycle. Therefore, we choose LTE-

U as the bearer technology for LTE-M service in

unlicensed spectrum.

The rest of this paper will analyse the short-

comings of origin CSAT when carrying LTE-M

service and propose a solution to improve the

performance of CSAT for LTE-M in unlicensed

spectrum. The paper is organized as follows. In

Section 2, we analyse the shortcomings of LTE-U

with CSAT mechanism to carry LTE-M service.

Then introduce the design of queue-sensing-based

CSAT in Section 3. And verify the performance of

improved CSAT by simulation in Section 4. Finally,

we conclude this work in Section 5.

2 SHORTCOMINGS OF CSAT

To guarantee the fairness of spectrum utilization in

unlicensed spectrum, LTE-U is supposed to adopt

CSAT mechanism. The rule for CSAT to adjust the

duty cycle is denoted as

,max 1

,min 2

(n 1) min(T (n) ,T ) (n)

(n 1) (n) (n)

(n 1) max( (n) ,T ) (n)

ON ON UP ON Thr

ON ON Thr Thr

ON ON DOWN ON Thr

T MU MU

T T MU MU MU

T T MU MU



    



   





    





(1)

where

(n)

is the transmission duration of

LTE-U in the

cycle, duration of which is

denoted as

CSAT



and



DOWN

are the steps to

increase and decrease

(n)

(n)MU

is the

estimation of Medium Utilization (MU) which

represents the channel occupancy of WLAN.

1Thr

and

2Thr

are the thresholds of MU to determine

the change of

 

1

,maxON

is a limit imposed on

the maximum time that

(n)

can take.

,minON

defined as follows based on the number of WLAN

Access Points (AP) that are operating in the

environment:

,min

(N 1) T

min sec,

CSAT

T TONMininmilli

N NumWiFiNodes













(2)

where

secTONMininmilli

is a configurable

parameter aimed at guaranteeing a minimum duty

cycle to avoid undetected APs’ transmission. N is

the number of LTE-U cells in the scenario.

NumWiFiNodes

is the number of detected APs.

Since the number of WLAN APs in the real

urban rail transit scenario is much larger than the

number of LTE-U cells, the calculation result of

,minON

is close to 0, which does not limit the

minimum duty cycle of LTE-U. Therefore, when

MU exceeds the threshold

2Thr

, the duty cycle of

LTE-U will continue to decline until reach

,minON

which will cause the security of LTE-M service

under threat. We will prove this by simulation in

Section 4. To avoid this happening, we improve

traditional CSAT with Frame Level Scheduler (FLS)

algorithm to ensure the reliable transmission of

LTE-M service.

3 DESIGN OF QUEUE-SENSING-

BASED CSAT

3.1 FLS Algorithm

FLS is a resource allocation algorithm for LTE

downlink which defines frame by frame the amount

of data that each service should transmit to satisfy its

delay constraint (Piro et al, 2011). The algorithm is

as follows.

where

 

is the quota of data that should transmit

in the

frame to meet its Quality of Service (QoS)

constraints.

 

is the queue length at time

which represents the starting time of the

frame.

 

is a system parameter to control the

distribution of data to transmit in each frame. It has

been proved that, for a service, if the amount of

transmitted data in the

frame can always meet

the constraint of

 

, a queuing delay smaller than

M+1 frame length can be provided.

3.2 System Model

FLS is a resource allocation strategy for discrete

events. It does not care about the change of queue

length between two samples, but only evaluate the

minimum amount of data to be transmitted in the

next sampling interval based on the queue length

obtained at sampling time. Therefore, change the

sampling interval does not affect the function of the

algorithm. Changing the sampling interval from a

frame to an LTE-U duty cycle length can easily

realize the combination of FLS and CSAT.

Figure 1 is the design of LTE-U transmission

process in downlink with queue-sensing-based

CSAT mechanism. In the figure, the solid line

represents the transmission process of data, and the

dotted line represents the work process of queue-

           

1 2 1

u k q k q k n q k n u k n c n



         







(3)

Queue-Sensing-Based CSAT: A Downlink Transmission Scheme for LTE-M in Unlicensed Spectrum

sensing-based CSAT, during which there is no data

transmitted.

Figure 1: Design of queue-sensing-based CSAT.

FLS scheduler first obtains the current queue

length

 

at each sampling time, according to

which it calculates and saves

 

. In the mean-

while, LTE-U transmitted data counter counts the

amount of data transmitted in last LTE-U cycle, and

save it as

 

1uk

. Then FLS scheduler comparing

 

1uk

with

 

1uk

, and adjust the duty cycle for

the current LTE-U cycle according to the

comparison result. The rule for queue-sensing-based

CSAT to adjust the duty cycle is denoted as

 

   

 

   

 

,max 1

,min 2

(n 1) min(T (n) ,T ) (n) 1 1

(n 1) (n) (n)

(n 1) max( (n) ,T ) (n) 1 1

ON ON UP ON Thr

ON ON Thr Thr

ON ON DOWN ON Thr

T MU MU u k u k

T T MU MU MU

T T MU MU u k u k



        



   





        





(5)

The rule introduces FLS based on CSAT.

Compared with the original algorithm, besides the

case where MU is smaller than the threshold

1Thr

 

1uk

is smaller than

 

1uk

, LTE-U will also

increase the transmission duration for the current

cycle to ensure that the transmission performance of

LTE-U meets the requirement of beared service.

Additionally, only if MU greater than the threshold

2Thr

and

 

1uk

is greater than

 

1uk

, will

LTE-U accept the reduction of duty cycle. Because

LTE-M service has a high requirement for security,

ensuring the reliable transmission of the service is a

priority. Therefore, only if the transmission perfor-

mance meets the requirement of LTE-M service,

will LTE-U accept to decrease its transmission

duration. Otherwise, the requirement of adjusting the

duty cycle will be refused.

Queue-sensing-based CSAT provides a way to

calculate the minimum amount of data that needs to

be transmitted during the current LTE-U cycle based

on the queue length. and make up for the lack of

attention to LTE-U performance in original CSAT.

With this solution, LTE-U could be a reliable bearer

for LTE-M service in unlicensed spectrum.

4 SIMULATION RESULTS

4.1 Simulation Scenario

Figure 2 is the coexistence scenario for LTE-U and

WLAN used in the simulation, which follows the

indoor scenario in 3GPP TR36.889 (3GPP, 2015). A

File Transfer Protocol (FTP) service is used for

transmission tests. The packet arrival rate of the

service obeys the Poisson distribution, and the

packet size is fixed to 0.5MB.

Figure 2: Coexistence scenario.

Figure 3: Result of transmission test.

Figure 3 is the simulation result of transmission

test in downlink of the LTE-U CSAT and WLAN

coexistence scenario mentioned above. In the test,

the beared service of LTE-U is set to be a stable data

stream with a data arrival rate of 10Mbps to simulate

LTE-M service. And by gradually increasing the

data arrival rate of WLAN, the overall throughput of

the channel can gradually approach to the saturation

capacity. The thresholds

1Thr

and

2Thr

are set

to be 0.2 and 0.4.

Figure 3 shows the curve of overall throughput

of the channel which tends to be flat when the data

arrival rate of WLAN reaches 120Mbps, where the

channel is closed to saturation. Therefore, in the

following simulation, the data arrival rates of

WLAN are selected to be 10Mbps, 60Mbps and

120Mbps to represent low, medium and high

channel load condition respectively.

CTISC 2019 - International Conference on Advances in Computer Technology, Information Science and Communications

4.2 Simulation for CSAT

As shown in Figure 3, the throughput of LTE-U with

CSAT mechanism decreases as the channel load

increases. This is because of the shortcomings of

CSAT mentioned in Section 2, when the channel

load is aggravated, the data transmission duration of

LTE-U will decrease and result in a reduction on

throughput as shown in Figure 4.

Figure 4: Duty cycle distribution under different load.

Figure 5: LTE-U throughput under different load.

Figure 6: LTE-U latency under different load.

Figure 4 is the distribution of LTE-U duty cycle

under different channel load, in which A, B, C

subgraphs represent the distribution under low,

medium and high channel load conditions respect-

tively. In the figure, when the channel load is low,

the distribution of duty cycle concentrates around

the value of 0.75, which is the maximum value of

duty cycle in the simulation. As the channel load

increases, the distribution begins to concentrate

around the minimum value. Finally, when the

channel load becomes to the high level, almost all

the duty cycle distributes at the minimum value.

Figure 5 and Figure 6 are the Cumulative

Distribution Probability (CDF) of the throughput

and latency of LTE-U CSAT under different channel

load. In the figure 5, as the channel load increases,

the throughput of LTE-U drops sharply. Under the

high channel load condition, the throughput is

almost zero. In the Figure 6, the distribution of delay

is consistent with which of throughput. Under the

high channel load conditions, the transmission delay

of LTE-U increases dramatically. Therefore, the

shortcomings of traditional CSAT mentioned in

Section 2 has been proved.

Due to the LTE-M specification, the service

delay should be less than 500ms, so LTE-U with

traditional CSAT cannot provide a reliable

transmission for LTE-M service.

4.3 Simulation for Queue-Sensing-

Based CSAT

4.3.1 Performance Verification

The cycle duration

CSAT

is set to 80ms. According to

the requirement of LTE-M service for the delay of

service not exceeding 500ms, the FLS parameter M

is set to 5, which means that the delay of LTE-M

service will not exceed (5+1)



80ms=480ms.

Figure 7 and Figure 8 are the CDF of the

throughput and latency of LTE-U with queue-

sensing-based CSAT under different channel load.

And the subgraphs A, B, C in both figures represent

the data arrival rates of LTE-U are set to 5Mbps,

10Mbps and 20Mbps respectively. Comparing

Figure 7 subgraph B with Figure 5 where the data

arrival rates of LTE-U are the same, under the low

channel load condition, two transmission schemes

have similar performance. This is because when

channel load is low, the transmission performance of

LTE-U can meet the requirement of LTE-M, and

MU is smaller than the threshold

1Thr

, so the

improved CSAT works as the original CSAT to

increase the transmission duration of LTE-U and get

the same performance. As the channel load increases,

the improved CSAT begins to show a better

performance than the original one, and under the

high channel load condition, the improved CSAT

can still keep a proper throughput for LTE-U to

ensure the reliable transmission of LTE-M service.

In the Figure 8 subgraph B, the distribution of

delay is consistent with which of throughput.

Significantly, the service delay under the high

Queue-Sensing-Based CSAT: A Downlink Transmission Scheme for LTE-M in Unlicensed Spectrum

(A) (B) (C)

Figure 7: Queue-sensing-based LTE-U throughput under different load.

(A) (B) (C)

Figure 8: Queue-sensing-based LTE-U latency under different load.

channel load condition is still less than 500ms. The

performance of improved CSAT has been verified.

4.3.2 Adaptive Verification

Comparing A, B, C subgraphs in Figure 7, we can

see that under the low channel load condition, as the

data arrival rate of LTE-U increases, the distribution

of throughput gradually moves towards a smaller

value. This is because when the channel load is low,

the duty cycle of LTE-U stays at the maximum value

for most of the time, so when the data arrival rate

increases, LTE-U cannot obtain more transmission

duration for extra data transmission, which causes

the queue length of LTE-U increasing, and the

performance of service delay and throughput getting

worse. On the contrary, when the channel load is

high, as the data arrival rate increases, the perfor-

mance of LTE-U throughput does not change a lot or

even better. This is because, under the high channel

load condition, the transmission duration of LTE-U

is compressed in a low level, and only require the

minimum resource to keep the normal operation of

LTE-M service. When the data arrival rate increases,

LTE-U can require more transmission duration

through the FLS scheduler to obtain even higher

instantaneous throughput, but the overall distribution

of throughput will not change greatly.

In figure 8, the distribution of delay is consistent

with throughput. When the data arrival rate of LTE-

U increases, individual packets have an increase in

delay, but all packets have a delay less than 500ms.

In conclusion, the queue-sensing-based CSAT can

make adaptive adjustments based on the change of

service requirement to ensure the reliable trans-

mission of beared service.

4.3.3 Fairness Verification

Figure 9 is the distribution of LTE-U duty cycle

under different channel load, and in subgraphs A, B,

C the data arrival rates of LTE-U are set to 5Mbps,

10Mbps and 20Mbps respectively. Comparing

Figure 9 subgraph B with Figure 4 where the data

arrival rates of LTE-U are the same, under the low

channel load condition, they have the similar

distribution of duty cycle as analysed above. As the

channel load increases, the distribution of duty cycle

at the minimum value in Figure 9 is less than which

in Figure 4. This is because improved CSAT

increases the duty cycle in some of LTE-U cycle to

obtain better transmission performance according to

the requirement of LTE-M service, but the overall

distribution still concentrates around the value of

0.05. Comparing subgraphs A, B and C, we can see

the similar distribution under the low channel load

condition. And under the medium or high channel

CTISC 2019 - International Conference on Advances in Computer Technology, Information Science and Communications

(A) (B) (C)

Figure 9: Queue-sensing-based LTE-U l duty cycle distribution under different load.

load condition, as the data arrival rate of LTE-U

increases, the distribution at the minimum duty cycle

gradually decreases, and the reduced portion are

distributed to other larger values. This is because the

greater the data arrival rate is, the more transmission

duration LTE-U needs, so the distribution changes

as above. The result above indicates that queue-

sensing-based CSAT adjust transmission duration of

LTE-U according to the requirement of beared

service. When the amount of data to transmit is large,

improved CSAT will allocate more duration for

LTE-U, and if the amount becomes smaller, it will

return the transmission duration to WLAN. This

mechanism maximizes the fairness of channel

occupancy in unlicensed spectrum under the premise

of ensuring the transmission performance of LTE-M

service.

5 CONCLUSIONS

In this paper, we propose a queue-sensing-based

CSAT downlink transmission scheme based on

LTE-U. The scheme provides a minimum amount of

data to transmit in the current transmission duration

as a measure of the requirement of LTE-U trans-

mission performance, according to which, LTE-U

dynamically adjusts its duty cycle to provide a

reliable transmission for LTE-M service in unli-

censed spectrum. According to the results of

simulation, this scheme also has a good performance

in the fairness of channel occupancy.

Furthermore, future investigations will focus on

the improvements on the duty cycle adjustment

strategies, and introduce more factors which affect

the transmission performance of LTE-U to get a

more stable transmission scheme which adapted to

various channel conditions.

ACKNOWLEDGEMENTS

This work is partly supported by Beijing Natural

Science Foundation (L161005).

REFERENCES

Beijing Radio Administration (2017) Notice of Frequency

for Beijing Rail Transit Video Transmission Wireless

Dedicated Network. Beijing: Beijing Radio

Administration.

China Urban Rail Transit Association (2016) Urban Rail

Transit Train-ground Integrated Communication

System (LTE-M) Specification. Beijing: China Urban

Rail Transit Association.

T., Shao and X., Xie (2017). ‘Construction of TD-LTE

Network for Hangzhou Subway Line 4’, Urban Fast

Track Traffic, 30 (3), pp. 102-106.

H., Hu, M., Zheng, K., Yu and B., Zhou (2016) ‘Enhanced

listen-before-talk scheme for LTE in unlicensed band’,

2016 6th International Conference on Electronics

Information and Emergency Communication

(ICEIEC). Beijing, 2016. pp. 168-173.

S., Choi and S., Park (2015) ‘Co-existence analysis of

duty cycle method with Wi-Fi in unlicensed bands’,

2015 International Conference on Information and

Communication Technology Convergence (ICTC).

Jeju, 2015. pp. 894-897.

Qualcomm (2014) LTE in Unlicensed Spectrum:

Harmonious Coexistence with Wi-Fi. San Diego, CA:

Qualcomm.

G., Piro, L.A., Grieco, G., Boggia, R. Fortuna and P.

Camarda (2011) ‘Two-Level Downlink Scheduling for

Real-Time Multimedia Services in LTE Networks’,

IEEE Transactions on Multimedia, 13 (5), pp. 1052-

1065. doi: 10.1109/TMM.2011.2152381.

3GPP (2015) Study on Licensed-Assisted Access to

Unlicensed Spectrum. Valbonne, France: 3GPP.

Queue-Sensing-Based CSAT: A Downlink Transmission Scheme for LTE-M in Unlicensed Spectrum