The Efficient Distribution Method of Limited Wireless

Communication Frequency Resources for the Multi-robot Teaming

Heeseo Chae

, Jae Hyuk Ju

and Jae Hyun Park

Defense Robotic Systems Team, Hanwha Defense, Pangyo-Ro, Bundang-Gu, Korea

Keywords: Wireless Communication Frequency, Multi-robot Teaming, Resource Distribution Method.

Abstract: In the situation where various defense robot systems are developed and operated, wireless network is an

indispensable element to remotely control unmanned robots. However, when each defense robot is operated

on the basis of wireless communication, the available frequency resources per robot are limited. At this point,

if multiple robot operations are increased and the number of robots participating in them increased

exponentially, a serious shortage of available frequencies is expected. Therefore, we propose a dynamic

allocation of frequency resources as the bottom-up type approach to overcome this problem. More specifically,

we implemented a bandwidth allocation scheme according to the priority change of each end node. We also

show the validity and efficiency of this method through the related experimental results.

1 INTRODUCTION

When an unmanned robot is operated at a remote

control station in real field as a defense use, a wireless

communication environment should be established

basically irrespective of whether the robot is a ground

vehicle or an aerial vehicle. However, the available

bandwidth of the communication frequency is limited,

and the number of the robot for the military purpose

is steadily increasing (Alberts, David S., et al., 2000).

In such a situation, when a plurality of robots are

operated by a single control system or heterogeneous

robots are operated together, a serious shortage of

wireless network frequency is expected. Especially,

military frequency is strictly controlled due to

security problems in military operation environment

(Heeseo Chae, et al., 2017). The unmanned robot we

are developing currently needs to be allocated a

communication frequency from the government in

order to use it in the actual field. However, there are

many difficulties.

Because not only new unmanned

robots but also existing systems use similar

communication equipment, the available frequency

bands are similar. Thus, users should pre-check and

avoid the preferred communication frequency band.

Depending on the area, they have to be preoccupy the

https://orcid.org/0000-0001-7293-5807

https://orcid.org/0000-0001-7547-9931

https://orcid.org/0000-0001-9664-086X

available frequencies and use it selectively.

(Jae Hyuk

Ju, 2014). Therefore, multi-robot operation method

and frequency utilization approach are being

researched to overcome this limited environment (D.

Gesbert, et al., 2003).

In this paper, we describe a case where

heterogeneous robots operating in a limited wireless

communication environment are teaming through

multiple robot control methods. In order to use

allocated frequencies effectively, we will distribute

them in accordance with each mission and suggest a

way to reallocate the network channels in a bottom-

up manner. We also show the system design and the

software architecture that will enable such

mechanism, and explain the implementations. Finally,

we analyse the experimental results applied to the

actual multi-robots and confirm the feasibility and

expandability of our approach.

2 CONFIGURATRION OF THE

DEFENSE ROBOT

There are three types of defense robots that we

develop. Size and weight are the criteria, and as

530

Chae, H., Ju, J. and Park, J.

The Efﬁcient Distribution Method of Limited Wireless Communication Frequency Resources for the Multi-robot Teaming.

DOI: 10.5220/0007971605300536

In Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2019), pages 530-536

ISBN: 978-989-758-380-3

shown in Figure1, it is classified into large-class,

medium-class, and small-class unmanned robots.

Even if the types of robots are different, each network

device used for wireless communication has a

common functions. There is some performance

difference depending on the operating range of the

robot (Hyo Keun Lee, et al., 2017). In the actual field,

the robot is remotely operated using the legitimate

network frequency assigned by each military unit. In

order to do teaming among the robots of different

classes, it is necessary to separate the channel within

the provided communication frequency range,

allocate them to the respective robots, and operate

them appropriately (Heeseo Chae, et al., 2016).

Figure 1: Multi-robot teaming concept and each robot spec.

The above figure shows the concept that each robot

operates remotely around the control station.

Although it operates in a top-down approach with in

the allocated frequency, it may also perform direct

communication between robots through the ad-hoc

network if necessary (Heeseo Chae, et al., 2018). In

addition, according to the driving ability and the role

of each robot, there is a difference in priority in using

network resources.

2.1 The Common Network Device Design

For multi-robot teaming, hardware and software of

communication devices should be provided as the

same common platform for each robot.

Figure 2: The Network device to set the dynamic

bandwidth.

In the past, when the frequency was allocated to each

device, the channel was distributed as the same

bandwidth (Albus, J. S. et al., 2006). This is a top-

down approach, and we use the dynamic bandwidth

configuration scheme shown in Figure 2 to improve

it. This method is the common design of the hardware

side, and in terms of software, the common communi-

cation architecture is required to implement it.

2.2 The Common Network Architecture

The runtime view of the network architecture, which

is mainly performed by the control station, is as

follows.

Figure 3: The Network architecture for server process.

The operator’s command are transmitted through the

process that is dedicated to each robot. Utilizing this

multi-thread structure, N robots for teaming can be

remotely controlled as long as communication

resources are provided. The heartbeat is periodically

The Efﬁcient Distribution Method of Limited Wireless Communication Frequency Resources for the Multi-robot Teaming

531

transmitted to prevent the network disconnection

between the control station and the robot. In order to

cope with dynamic bandwidth change, it is possible

to continuously identify the robot that was being

controlled by using a key for each robot.

3 FREQUENCY DIFFERENTIAL

DISTRIBUTION SCHEME

If a large number of unmanned robots are operated

according to the conventional method within a limited

frequency, it is inevitable to provide insufficient

bandwidth to all robots. That is, the legacy approach

using the even distribution can only operate a limited

number of robots, and is unsuitable for operating

more than N robots. Therefore, the proposed

frequency operation method is dynamically

determined without setting the division ratio in

advance.

3.1 Resource Allocation by the Robot

Teaming Scale

The method of distributing the bandwidth according

to the scale of robot teaming is as follows.

Figure 4: The channel allocation by the teaming scale.

In the figure 4, case#1 and case#2 are examples of

frequency allocation using even distribution. When

the number of robots is small, remote control is

possible through real time image. However, as the

number of robot increase, it becomes difficult to

provide real time image. In the case#3, the master of

the robot team monitors the current task and network

status of each member, and send a channel

reallocation request to the control station at a

predetermined time interval. It can accept this

bottom-up request and provide new bandwidth

dynamically. When sharing information among

robots within a robot team, some portion of the

existing frequency bands can be used. However, in

order to maximize available frequency range, the

separated network as ad-hoc type is utilized. This area

uses a temporary frequency domain because it is a

short distance communication that exchanges simple

data such as status information of each robot. In order

to prevent attenuation by distance, ad-hoc needs to

maintain the distance between robots.

3.2 Network Channel Reallocation

Process and Criteria

Reallocating the wireless network bandwidth as a

bottom-up approach requires a formal process and

standard.

Figure 5: The recursive process for bandwidth reallocation.

In the above figure, except for the initialization

process (stage1 and stage2), it shows the cycle of

continuous circulation according to a certain time

period. The factors considered in channel reallocation

are the priority of the task being executed and the

urgency of the situation. This information is provided

to nearby robots and the master. Based on the data,

the bandwidth allocation is determined.

The criteria of the channel reallocation process is

based on the task priority table. It is as follows.

Stage1. Checking the number of operating robots

Stage2. Allocating default channel bandwidth

Stage3. Using the provided frequency during a period

Stage4. Prioritizing robot based on Mission & Situation

Stage5. Reallocating channel bandwidth

Stage6. Recursive call

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

532

Table 1: Task Priority Matrix from the Operation Concept.

Driving

Status

Task for Mission

Surveillance firing No mission

Remote

Driving

2 1 3

Autonomous

Driving

5 4 6

Stop 8 7 9

The missions of robots are simplified as surveillance,

shooting, and moving. The highest priority is

shooting. This is because the target must be captured

with real-time video and fire commands should be

delivered without delay. In case of driving, remote

driving and autonomous driving are classified. The

remote driving requires a large amount of

communication bandwidth because it needs real-time

video streaming. Disconnection of network during the

driving cab lead to accidents. On the other hand,

autonomous driving requires relatively fewer videos,

so it can be operated with a small traffic volume

(DeSouza, Guilherme N., et al., 2002). Through a

combination of mission and driving, the situation of

firing during remote driving requires the most

resources. Conversely, if it does nothing and stops, it

has the lowest priority. In this case, the network

bandwidth is reduced dynamically.

3.3 Frequency Reuse Considering

Interference

If a multi-robot team is operated by the proposed

method, interference of the network frequency can be

predicted considering the operation range of each

team. It is shown in the figure below.

Figure 6: Frequency Reuse Situation based on the distance.

Because there is no interference between Robot

Team#1 and Robot Team#2, frequency reuse is

possible. The communication radius between the

robot and the control station is 5km, which is beyond

the supportable range of the network device. There is

no problem to use the same frequency in different

teams. However, when an operation is performed at a

close distance, such as the Robot Team#2 and the

Robot Team#3, some robots can be overlapped. In

this case, frequency interference occurs and noise is

generated. To overcome this situation, we set a

unique OP-Code for communication network of each

team and set a robot key for individual end node. The

OP-Code can be generated by the Random Number

Generator with AES256 standard (Keller, Sharon S,

2005). Because of this function, even if the

communication is interrupted, unauthorized control

of the robot is impossible. This concept can be

achieved through a network process design that is

adaptable to dynamic bandwidth changes and a

packet structure with security considerations.

3.4 Packet Structure for Dynamic

Channel Change

The above-referred security-enhanced packet

structure can be seen in Figure 7 as a runtime view.

Especially, the design focused on the function of the

main console server of the control station enables to

cope with the changes of the robot’s mission and

driving status.

Figure 7: Dynamic bandwidth process and packet structure.

The network structure is implemented to prevent the

robot from being lost due to operator’s miss-control,

network disconnection, and frequency interference.

The robot and each manipulation device are not

directly connected, but data is exchanged through the

server. Also, the authority of a massage packet is

changed according to whether the current driving

state of the robot is remote driving or autonomous

5km

No Frequency

Interference

5km

Robot Team#1

5km

Frequency

Interference

Robot Team#3

Robot Team#2

The Efﬁcient Distribution Method of Limited Wireless Communication Frequency Resources for the Multi-robot Teaming

533

driving. With this design, the authority and the

security can be maintained when allocating dynamic

bandwidth. Such a mechanism can provide a wireless

communication environment in which reliability and

security are considered.

4 EXPERIMENT

In order to verify the proposed frequency distribution

scheme, the following comparative experiments were

conducted. One control station and four network

devices are selected, and a total of five network

devices are utilized.

4.1 Experimental Conditions

For the convenience of experiment, we used network

devices of two mobile robots and three stationary

terminals.

Figure 8: Experimental equipment and conditions.

The experimental site used the terrain with the road

side and the ground side. The network device has the

1W output and 4x4 MIMO type antennas. Also, the

frequency is 2.2~2.5GHz and the bandwidth is

basically 20MHz.

In the first experiment, we measured the transmission

performance by distance through the static channel

distribution. The second experiment, which we

suggested, we confirmed how efficiently the

transmission performance can be improved through

the dynamic channel distribution.

4.2 Result of Static Channel Distribution

When the control station was set as the origin, the

bandwidth was distributed to the same 5MHz for each

channel of each node. We measured the signal

sensitivity (SNR: Signal to Noise Ratio) by increasing

the distance of the nodes.

Table 2: Signal Sensitivity of Static Channel Distribution.

Distance

Node’s bandwidth

MUGV

[5MHz]

UCV

[5MHz]

ST#1

[5MHz]

ST#2

[5MHz]

0m 52dB 58dB 69dB 65dB

500m 35dB 42dB 50dB 44dB

1000m 30dB 35dB 42dB 34dB

1500m 22dB 34dB 39dB 34dB

2000m 20dB 30dB 32dB 33dB

2500m 17dB 24dB 32dB 26dB

3000m 15dB 20dB 23dB 20dB

3500m 8dB 11dB 26dB 21dB

4000m 11dB 18dB 23dB 21dB

4500m 15dB 17dB 28dB 21dB

5000m 11dB 12dB 29dB 21dB

When the signal sensitivity is 30dB or more, 5MHz

can be used as a full condition. Between 20dB and

30dB, 3Mbps can be used. Less than 20dB can be

used below 1Mbps, so it is difficult to transmit video

in real-time. It can be expressed as follows.

Figure 9: Throughput of Static Channel Distribution.

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

MUGV UCV ST#1 ST#2

(m)

(Mbps)

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

534

There is some performance difference between anten-

nas, but communication sensitivity is further reduced

in the case of moving robots. It limits the distance

remotely controllable. The stationary terminals do not

need to move, so the network sensitivity does not need

to be good. However, since they exhibit better signal

sensitivity, waste of resources occurs.

4.3 Result of Dynamic Channel

Distribution

The bandwidth of each node is differently distributed.

By increasing the priority of the mobile robots, more

bandwidth was allocated to MUGV (Multi-purpose

UGV) and UCV (Unmanned Combat Vehicle) by

8MHz, respectively. Likewise, we measured the

signal sensitivity by increasing the distance from the

origin (Control Station). Unlike the first experiment,

if the SNR is over 30dB in 8MHz node, 8Mbps can

be used. If the SNR is less than 20dB in 8MHz node,

still it can use 4Mbps. However, a node that has been

allocated a 2MHz bandwidth can only use 2Mbps at

30dB. The table shows the following.

Table 3: Throughput table by SNR and Bandwidth.

SNR

Node

MUGV

[8MHz]

UCV

[8MHz]

ST#1

[2MHz]

ST#2

[2MHz]

30dB~ 8Mbps 8Mbps 2Mbps 2Mbps

20~30dB 6Mbps 6Mbps 1.2Mbps 1.2Mbps

~20dB 4Mbps 4Mbps 0.5Mbps 0.6Mbps

The available range of communication derived from

the above table is as follows.

Table 4: Signal Sensitivity of Dynamic Channel Distribution.

Distance

Node’s bandwidth

MUGV

[8MHz]

UCV

[8MHz]

ST#1

[2MHz]

ST#2

[2MHz]

0m 52dB 58dB 67dB 62dB

500m 33dB 41dB 52dB 45dB

1000m 31dB 36dB 43dB 33dB

1500m 22dB 35dB 37dB 31dB

2000m 21dB 32dB 28dB 30dB

2500m 18dB 33dB 29dB 24dB

3000m 16dB 23dB 23dB 20dB

3500m 11dB 12dB 25dB 19dB

4000m 10dB 13dB 24dB 19dB

4500m 15dB 15dB 28dB 21dB

5000m 11dB 13dB 27dB 21dB

The sensitivity of the signal along the distance is

similar to the first experiment. However, even though

the SNR of the mobile robot decrease according to the

distance, it can be confirmed that the throughput at

the long distance is more than 4Mbps because the

allocated resources are large. On the other hand, the

stationary terminal does not have a small SNR, but

the allocated bandwidth is so small that the

throughput is not enough to transmit videos even at

close distance. It is shown in the graph below.

Figure 10: Throughput of Dynamic Channel Distribution.

5 CONCLUSIONS

Since a large number of defense robots are expected

to be operated in the near future, the method for

efficiently using a limited network frequency has

suggested. The communication bandwidth was

reallocated on the basis of the mission and the driving

situation of the robots. As a result, we overcome the

shortage of network resource as much as possible. We

also proposed a network architecture and a packet

structure that can be applied irrespective of frequency

interference. The mechanism was verified by two

comparative experiments such as the static and the

dynamic channel distribution. If you use the dynamic

channel distribution to provide a significant amount

of resources to robots that require them, it is

experimentally confirmed that it is possible to

communicate smoothly even at a distance of about

two to three times as compared with the existing one.

In the future, more robots will be operated to

confirm the scalability of the proposed method. We

will also complete a mathematical model that takes

into account various optimization parameters such as

climate, terrain, jamming, increasing hop, and

moving speed of the robot. Through this, it is expect

to improve the reliability of the proposed method by

testing multi-robots with more than four nodes in the

virtual environment that could not be performed in

the field experiment.

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

MUGV UCV ST#1 ST#2

(m)

(Mbps)

The Efﬁcient Distribution Method of Limited Wireless Communication Frequency Resources for the Multi-robot Teaming

535

ACKNOWLEDGEMENTS

This research was carried out with the support of the

civilian military applied research by Hanwha

Defense.

REFERENCES

Alberts, David S., et al. " Network Centric Warfare:

Developing and Leveraging Information Superiority”,

Assistant Secretary of Defense (C3I/Command Control

Research Program) Washington DC, 2000.

Heeseo Chae, et al. "The Reliable Design by the Wireless

Network Redundancy for Remote Control Mechanism

of the Heavy Unmanned Ground Vehicle", AIM2017,

IEEE, 2017.

Jae Hyuk Ju, " Automatic Traffic Control for Robot”, KR

Patent 2014-86157, 2014.

D. Gesbert, et al. "From theory to practice: an Overview of

MIMO space-time coded wireless systems”, IEEE

Journal on Selected Areas in Communications, vol. 21,

no. 3, pp. 281-302, 2003.

Hyo Keun Lee, et al, “A Trend Analysis and Technology

Application of Defense Technology Protection”,

Journal of Kimst, vol.20, Issue 4, p579-586, 2017.

Heeseo Chae, et al. “The comparison of the detecting

Performance between the ground and the aerial visual

analytics in the UGV-UAV collaborative system”, IEEE

conference (AIM), 2016

Heeseo Chae, et al. "The Anti-tampering process and case

study by the operating mode of various unmanned

ground vehicle”, AIM 2018, IEEE, 2018.

Albus, J.S. et al. " Learning in a Hierarchical Control

System: 4D/RCS in the DARPA LAGR Program”,

ICINCO 06, 2006.

DeSouza, Guilherme N., et al. " Vision for mobile robot

navigation: A survey”, IEEE transactions on pattern

analysis and machine intelligence 24.2, 2002.

Keller, Sharon S. “NIST-recommended random number

generator based on ANSI X9. 31 appendix A. 2.4 using

the 3-key triple DES and AES algorithms”, NIST

Information Technology Laboratory-Computer

Security Division, National Institute of Standards and

Technology, 2005.

ICINCO 2019 - 16th International Conference on Informatics in Control, Automation and Robotics

536