A Comparison among Wi-Fi Direct, Classic Bluetooth, and Bluetooth

Low Energy Discovery Procedures for Enabling Massive Machine Type

Communications

Abel Rodriguez Medel and Jose M. Camara Brito

National Institute of Telecommunications INATEL, Santa Rita do Sapucai, MG, Brazil

Keywords:

IoT, mMTC, D2D.

Abstract:

The exponential growth of the Internet of Things (IoT) devices bring the necessity to support massive Ma-

chine Type Communications (mMTC) for the Next-Generation networks. One of the enablers for mMTC is

the Device-to-Device (D2D) communications. Since it is not possible to reduce the number of devices in mas-

sive communications, all the effort for collision and power consumption reduction should be applied to the

discovery algorithm of the D2D technologies. The principal D2D technologies are Wi-Fi Direct, Classic Blue-

tooth, and Bluetooth Low Energy (BLE). However, most of these technologies were not originally designed

to support massive communications. The main goal of this work is to assess Wi-Fi Direct, Classic Bluetooth,

and BLE performances in terms of number of collisions, energy consumption, and discovery latency, in order

to check out the more suitable technology for mMTC scenarios. The results show that Classic Bluetooth is

faster than Wi-Fi Direct during the devices’ discovery, accelerating network access for the devices in massive

communications. Besides, BLE incurs fewer collisions, less energy consumption, and less time for devices’

discovery than Classic Bluetooth. Thus, BLE is the more suitable D2D technology—out of the three analyzed

in this work—to enable mMTC.

1 INTRODUCTION

IoT is a new technology paradigm capable of offering

data analytics and insights, prediction models, and re-

mote monitoring and surveillance. IoT enables count-

less applications like forecasting upcoming network

behaviors through the data retrieved by pools of sen-

sors; establish energy-efﬁcient machine-to-machine

communications to achieve low power Wide Area

Networks (WAN); enable device-to-device communi-

cations to share network resources and locate nearby

devices; empower autonomous vehicles communica-

tions between cars and with the road infrastructure;

monitor equipment health and performance, and pro-

tecting physical infrastructure. Therefore, IoT has

emerged as the principal enabler for the majority of

the large-scale services that will be offered by future

networks. All these services require a great number of

interconnected devices to collect the big data used for

the processing. The higher the number of devices, the

higher the network’s resources to offer access to ev-

ery device. However, the current networks by them-

self have limited resources and they cannot attend

the millions of devices. Besides, the IoT devices are

resource-constrained especially in terms of energy.

Therefore, D2D communications have arisen to efﬁ-

ciently manage the mobile network’s resources and

overcome link budget problems for ultra-low power

devices (Ali et al., 2015).

The D2D communications include short-range

wireless technologies like Wi-Fi Direct, Classic Blue-

tooth, and Bluetooth Low Energy (Ali et al., 2015).

In all of them is present a discovery procedure where

devices ﬁnd each other to establish a communication

in a speciﬁc frequency of the spectrum. During the

discovery procedure, every device has two possible

roles: inquirer or scanner. An inquiring device peri-

odically sends signals in different frequencies to ad-

vertise nearby devices that an inquiring device is look-

ing for some peers to establish communication. A

scanning device periodically listens to the signals sent

by the device executing the discovery. The scanning

device then sends back another signal to let the inquir-

ing device know that the scanning device is available

to communicate in the frequency the inquiring device

is transmitting the discovery signal. The only way

the devices communicate is that the device executing

the inquiring task and the device scanning for incom-

164

Medel, A. and Brito, J.

A Comparison among Wi-Fi Direct, Classic Bluetooth, and Bluetooth Low Energy Discovery Procedures for Enabling Massive Machine Type Communications.

DOI: 10.5220/0010399001640169

In Proceedings of the 6th International Conference on Internet of Things, Big Data and Security (IoTBDS 2021), pages 164-169

ISBN: 978-989-758-504-3

ing discovery signals transmit and receive the discov-

ery signal in the same frequency, respectively. How-

ever, the transmission of the discovery signal and the

reception of the acknowledge signal are conditioned

by other discovery procedures. The more discovery

procedures happening simultaneously in the same lo-

cation, the greater the collision probability and more

discovery delay. Therefore, the D2D technology em-

ployed in massive communications should address the

great number of collisions and high-power consump-

tion experienced by the devices. This paper describes

the discovery algorithms of the principal D2D tech-

nologies and compares the performances of Wi-Fi Di-

rect, Classic Bluetooth, and BLE, in order to select

the more suitable D2D technology to enable massive

communications.

The rest of this paper is organized as follows. Sec-

tion 2 describes the discovery procedures of the prin-

cipal D2D technologies. In section 3, the Classic

Bluetooth and Wi-Fi Direct discovery performances

are compared by analyzing the behavior of the two

technologies with a developed app. Section 4 de-

scribes the details of an additional back-off imple-

mented for Classic Bluetooth to reduce both the num-

ber of collisions and the elapsed times during devices’

discovery in mMTC scenarios. In section 5, the per-

formances of Classic Bluetooth and BLE are com-

pared by simulating their discovery procedures. Then,

section 6 concludes the paper.

2 DISCOVERY PROCEDURES OF

THE PRINCIPAL D2D

TECHNOLOGIES

Since the discovery procedure of the D2D tech-

nologies concerns a lot of signaling, it captures all

the attention, especially in massive communications.

Therefore, this section describes the algorithms em-

ployed by every D2D technology during the discovery

procedure.

In the case of Wi-Fi Direct, every device has two

states: searching state and listening state. Both states

have a random duration of N time units (102.4 ms),

where N ∈ {1, 2, 3, ...}. Figure 1 shows the discov-

ery procedure for Wi-Fi Direct. In the searching state,

devices broadcast probe requests (discovery requests)

in one of the three social channels: 1, 6, 11 in the

2.4 GHz band (Camps-Mur et al., 2013)(Khan et al.,

2017). In the same searching interval, the device lis-

tens to probe request replies. In the listening state,

devices only listen to probe requests in one of the so-

cial channels and send back responses in the corre-

sponding cases. The selected channel in the listening

state remains constant during the entire discovery pro-

cess. In this state, devices do not listen to responses of

their own past probe transmissions (Sun et al., 2016).

In conclusion, a device only discovers a remote de-

vice when it receives probe request responses in the

searching state.

Figure 1: Wi-Fi Direct discovery procedure between two

devices (Camps-Mur et al., 2013).

For Bluetooth, the device that starts discovering

nearby devices is called an inquiring device. It broad-

casts inquiry packets in 32 of 79 possible frequencies.

The 32 frequencies are previously agreed upon. The

inquiring device keeps sending inquiry packets for

two time-slots of 312.5 µs in two different frequen-

cies generated by an internal 28-bit clock, as is shown

in Figure 2. The device then listens in the next two

subsequent time slots in the same frequencies it sent

inquiry packets before. After the listening interval, if

the inquiring device does not receive a reply to the in-

quiry packets, it starts the inquiry packets’ broadcast

again in two other frequencies. The scanning device

uses the 28-bit clock to generate the frequency that the

scanning device will use to listen to inquiry packets.

The timing used by the scanning device is depicted in

Figure 3 (Duﬂot et al., 2006).

Figure 2: Classic Bluetooth inquiring device’s behavior

(Duﬂot et al., 2006).

Figure 3: Classic Bluetooth scanning device’s behavior

(Duﬂot et al., 2006).

In the case of BLE, there are only three channels

used for the discovery procedure. These channels are

A Comparison among Wi-Fi Direct, Classic Bluetooth, and Bluetooth Low Energy Discovery Procedures for Enabling Massive Machine

Type Communications

165

the 37, 38, and 39. Figure 4 (a, b) summarizes the

BLE discovery procedure. On the one hand (Fig-

ure 4a), an advertising device sends advertising PDUs

over the three channels during an Advertising Event.

Between two consecutive advertising events, there is

a variable Ta time, composed by a ﬁxed advInterval

and a pseudo-random advDelay. On the other hand

(Figure 4b), a scanning device periodically scans the

same three channels to look for advertising signals

during a scanWindow, which is within a scanInter-

val. In every scanWindow, the scanning device scans

a different channel from the three channels. The BLE

standard states that the advInterval should be an in-

teger multiple of 0.625 ms in the range of 20 ms to

10.24 s, the advDelay should be within the range of

0 ms to 10 ms, and the scanInterval and scanWindow

shall be less than or equal to 10.24 s (Liu et al., 2012)

(Cho et al., 2016).

Figure 4: BLE discovery procedure: a) Advertising process

and b) Scanning process (Liu et al., 2012).

3 CLASSIC BLUETOOTH AND

WI-FI DIRECT DISCOVERY

PERFORMANCES

COMPARISON

We developed two Android apps to test the Clas-

sic Bluetooth and Wi-Fi Direct elapsed times during

the devices’ discovery and incoming data processing.

The apps were tested on the same two devices: Sam-

sung Galaxy S8+ cell phone and a Huawei P9 cell

phone. After ten measurements, which is sufﬁcient

for the obvious difference in results, the average dis-

covery time for Wi-Fi Direct was 8164 ms. This re-

sult is almost ﬁve times greater than the Classic Blue-

tooth discovery time, as shown in Table 1. Therefore,

Classic Bluetooth is best suitable to be employed as

the D2D technology for enabling mMTC scenarios.

However, the Classic Bluetooth discovery procedure

can be applied to Wi-Fi to obtain a larger communica-

tion range. This means that the Classic Bluetooth dis-

covery algorithm would be used but using the trans-

mission power of Wi-Fi.

The two apps used for Classic Bluetooth and Wi-

Fi Direct in this comparison were developed in Qt

Creator and Android Studio using the Bluetooth and

Wi-Fi Application Programming Interfaces (APIs),

respectively. The apps can be found here https://

github.com/Abel1027/D2D-Test-Apps.git.

Table 1: Elapsed times in milliseconds for Classic Blue-

tooth and Wi-Fi Direct to ﬁnd a remote device in short-

range in 10 attempts.

Attempts Bluetooth (ms) Wi-Fi Direct (ms)

1 649 8346

2 2659 8630

3 1600 6405

4 898 6335

5 2137 13176

6 4366 7329

7 530 6176

8 3196 9630

9 823 6000

10 373 9613

Mean 1723.1 8164

Std 1269.16 2130.51

4 PROPOSED MODIFICATIONS

FOR THE CLASSIC

BLUETOOTH DISCOVERY

PROCEDURE

Since Classic Bluetooth has a shorter elapsed time

when discovering nearby devices in comparison to

Wi-Fi Direct, Bluetooth is studied for improvement to

apply for the discovery procedure in mMTC scenar-

ios. It was simulated an environment where a set of

devices (inquiring) tries to ﬁnd another set of devices

on discoverable mode (scanning). In Classic Blue-

tooth, if more than one device wants to discover other

devices and perform the inquiring task and start in-

quiring with the same sequence of frequencies, their

transmissions will collide in every attempt, and they

never will ﬁnd a remote device. Therefore, a back-

off is proposed to be added every 11.25 ms (the time

IoTBDS 2021 - 6th International Conference on Internet of Things, Big Data and Security

166

needed by a scanning device to listen to incoming dis-

covery messages). The back-off is computed as a ran-

dom value from [0, 1, . . . , 10]×312.5 mus where the

maximum value is 10×312.5 mus = 3.125 ms. The

proposed back-off avoids transmissions at the same

moment in the same frequency for a discrete simu-

lated environment with a resolution of 31.25 mus and

a communication latency of 0 ms. In real scenarios,

the back-off needs to be calculated depending on the

expected latency.

5 PERFORMANCE

COMPARISON BETWEEN

CLASSIC BLUETOOTH AND

BLE DISCOVERY

PROCEDURES

We simulated the behavior of the Classic Blue-

tooth discovery algorithm with the modiﬁcations pro-

posed in this work and the BLE discovery proce-

dure. The simulation was deployed by SimPy, a

process-based discrete-event simulation Framework

used through the Python programming language, us-

ing an MSI laptop with a Core i7 processor, a 16

GB RAM, and a 16 GB NVIDIA GeForce RTX 2070

video card. The scripts used for the simulation can

be found here: https://github.com/Abel1027/Classic-

Bluetooth-vs-BLE.git. The results of the simulation

are shown below after computing the average of 10

simulations by variating the random seed in the range

0-9.

Figures 5-8 show the simulation when the number

of inquiring devices augments from 10 to 100 in steps

of 10 devices, and the number of scanning devices re-

mains the same, in this case, 10 devices. Figures 9-12

show the simulation when the number of inquiring de-

vices remains the same (10 devices), and the number

of scanning devices augments from 10 to 100 in steps

of 10 devices. The x-axis of the ﬁgures shows the

number of inquiring/scanning devices. For example,

10/20 means that there are 10 inquiring devices and

20 scanning devices.

Figure 5 shows fewer collisions during the BLE

discovery procedure than the Classic Bluetooth dis-

covery procedure. Therefore, the energy consumption

and the discovery time are expected to be lower in the

case of BLE than in Classic Bluetooth.

The energy consumed by the devices is computed

as:

E =

P × m

(1)

where E is the consumed energy in every trans-

mission given in J, P is the transmission power in mW,

m is the number of bits of the transmitted message,

and R

is the bit rate.

The energy consumed by the inquiring devices

is calculated from Eq.(1) with P=6.31 mW, m=68

bits, and R

=1 Mb/s for Classic Bluetooth; and with

P=6.31 mW, m=128 bits ⇒ 8 bits (Preamble) + 32

bits (Access Address) + 64 bits (Connectable Undi-

rected Advertising packet) + 24 bits (Cyclic Redun-

dancy Check - CRC), and R

=1 Mb/s for BLE. The

energy consumed by the scanning devices is calcu-

lated from Eq.(1) with P=6.31 mW, m=286 bits ⇒

72 bits (Access Code) + 54 bits (Header) + 144 bits

(Payload) + 16 bits (CRC), and R

=1 Mb/s for Clas-

sic Bluetooth; and with P=6.31 mW, m=176 bits ⇒ 8

bits (Preamble) + 32 bits (Access Address) + 112 bits

(Connectable Directed Advertising packet) + 24 bits

(CRC), and R

=1 Mb/s for BLE.

From Figure 6, the energy consumed by the in-

quiring devices is less for the BLE case than for the

Classic Bluetooth case. The same occurs for the en-

ergy consumption of the scanning devices (see Figure

7).

Figure 8 shows that the total time elapsed during

the discovery procedure is less for the BLE case than

for the Classic Bluetooth case. In the BLE case, the

discovery time is below 10 seconds while the discov-

ery time for the Classic Bluetooth case is above 10

seconds.

From Figure 9, there are fewer collisions during

the BLE discovery procedure than the Classic Blue-

tooth discovery procedure. Therefore, the energy con-

sumption and the discovery time are expected to be

lower in the case of BLE than in Classic Bluetooth.

The energy consumed by the inquiring devices is

less for the BLE case than for the Classic Bluetooth

case, as it is depicted in Figure 10. The same occurs

for the energy consumption of the scanning devices

(see Figure 11).

Figure 12 shows that the total time elapsed during

the discovery procedure is less for the BLE case than

for the Classic Bluetooth case. In the BLE case, the

discovery time is below 1 second when the number

of scanning devices is greater than 20 while the dis-

covery time for the Classic Bluetooth case is always

above 10 seconds.

6 CONCLUSIONS

The D2D communication technologies have arisen

as the enablers of the massive communications for

the Next-Generation networks. Most of the available

A Comparison among Wi-Fi Direct, Classic Bluetooth, and Bluetooth Low Energy Discovery Procedures for Enabling Massive Machine

Type Communications

167

Figure 5: Total number of collisions during the discovery

procedure when the number of inquiring devices increases

and the number of scanning devices remains the same.

Figure 6: Total energy spent by the inquiring devices during

the discovery procedure when the number of inquiring de-

vices increases and the number of scanning devices remains

the same.

Figure 7: Total energy spent by the scanning devices during

the discovery procedure when the number of inquiring de-

vices increases and the number of scanning devices remains

the same.

Figure 8: Total elapsed time for all inquiring devices to ﬁnd

a scanning device during the discovery procedure when the

number of inquiring devices increases and the number of

scanning devices remains the same.

Figure 9: Total number of collisions during the discovery

procedure when the number of inquiring devices remains

the same and the number of scanning devices increases.

Figure 10: Total energy spent by the inquiring devices dur-

ing the discovery procedure when the number of inquiring

devices remains the same and the number of scanning de-

vices increases.

IoTBDS 2021 - 6th International Conference on Internet of Things, Big Data and Security

168

Figure 11: Total energy spent by the scanning devices dur-

ing the discovery procedure when the number of inquiring

devices remains the same and the number of scanning de-

vices increases.

Figure 12: Total elapsed time for all inquiring devices to

ﬁnd a scanning device during the discovery procedure when

the number of inquiring devices remains the same and the

number of scanning devices increases.

D2D technologies were not originally conceived to

support such a massiveness of devices trying to con-

nect with each other. Other D2D technologies were

born to handle mMTC scenarios, but they are not stan-

dardized yet. Therefore, it is useful to analyze, test,

and assess the available D2D technologies to check

out the technology with the best performance during

the discovery procedure, which is the process with

more signaling.

In this work we compared the discovery pro-

cedures’ performances of Wi-Fi Direct and Clas-

sic Bluetooth. Then, we compared Classic Blue-

tooth with BLE. The results show that Classic Blue-

tooth was 5 times faster than Wi-Fi Direct during the

devices’ discovery. However, Classic Bluetooth is

slower than BLE. BLE also experiences fewer col-

lisions and consumes less energy than Classic Blue-

tooth during the devices’ discovery. BLE allows that

around 100 inquiring devices ﬁnd at least 1 scanning

device out of 10 scanning devices in less than 10 sec-

onds. BLE also allows that around 10 inquiring de-

vices discover at least 1 scanning device out of 100

scanning devices in less than 1 second, in contrast

with the 10 seconds the Classic Bluetooth takes to

do the same task. Therefore, BLE is faster and con-

sumes less energy than Bluetooth, and Wi-Fi by tran-

sitivity. For these reasons, BLE is one of the most

promising D2D communication technologies—out of

the three analyzed in this work—for enabling mMTC

in the Next-Generation networks. However, BLE is

a very short-range D2D technology, and it has some

security vulnerabilities during the pairing procedure

which can be overcome by other D2D technologies

like LTE-Direct.

This work was partially supported by

RNP, with resources from MCTIC, Grant No.

01250.075413/2018-04, under the Radiocommuni-

cation Reference Center (Centro de Refer

encia em

Radiocomunicac¸

oes - CRR) project of the National

Institute of Telecommunications (Instituto Nacional

de Telecomunicac¸

oes - Inatel), Brazil.

REFERENCES

Ali, A., Hamouda, W., and Uysal, M. (2015). Next gener-

ation m2m cellular networks: challenges and practi-

cal considerations. IEEE Communications Magazine,

53(9):18–24.

Camps-Mur, D., Garcia-Saavedra, A., and Serrano, P.

(2013). Device-to-device communications with wi-ﬁ

direct: overview and experimentation. IEEE wireless

communications, 20(3):96–104.

Cho, K., Park, G., Cho, W., Seo, J., and Han, K. (2016).

Performance analysis of device discovery of bluetooth

low energy (ble) networks. Computer Communica-

tions, 81:72–85.

Duﬂot, M., Kwiatkowska, M., Norman, G., and Parker, D.

(2006). A formal analysis of bluetooth device discov-

ery. International journal on software tools for tech-

nology transfer, 8(6):621–632.

Khan, M. A., Ch

erif, W., Filali, F., and Ridha, H. (2017).

Wi-ﬁ direct research - current status and future per-

spectives. Journal of Network and Computer Applica-

tions, 93.

Liu, J., Chen, C., and Ma, Y. (2012). Modeling and perfor-

mance analysis of device discovery in bluetooth low

energy networks. In 2012 IEEE Global Communica-

tions Conference (GLOBECOM), pages 1538–1543.

IEEE.

Sun, W., Yang, C., Jin, S., and Choi, S. (2016). Listen chan-

nel randomization for faster wi-ﬁ direct device discov-

ery. In IEEE INFOCOM 2016-The 35th Annual IEEE

International Conference on Computer Communica-

tions, pages 1–9. IEEE.

A Comparison among Wi-Fi Direct, Classic Bluetooth, and Bluetooth Low Energy Discovery Procedures for Enabling Massive Machine

Type Communications

169