• 検索結果がありません。

Swarm of computational clouds control system

Communication Control and Network

4.2 Swarm of computational clouds control system

Chapter 4

Communication Control and

Table 4.1:Communication control and network

Problem Details of the issue Proposed Solution

The overload and defi-ciency of the GCS af-ter controlling multiple UAV via an IoT service.

When the GCS has installed in the cloud server, The cloud cannot handle multi-UAV effi-ciently.

A swarm of computational clouds is integrated to enhance the GCS performance when multi-UAV is used.

The master protocol is able to use two different types of protocols when using IoT.

Evaluation of TCP and UDP protocols was missing and used randomly in previous re-searches.

TCP and UDP protocols have been evaluated in real flight experiment, and an evaluation has been given.

Heterogeneous network environment.

A heterogeneous network is a possible scenario in a disaster scenario. There is no technical solution to support UAV com-munication system in a het-erogeneous network environ-ment.

NTMobile is integrated into UAV and GCS and evaluated through real flight experiment where UAV and remote con-trol device have been forced to exchange networks.

In previous work [Aljehani, 2016d] in Chapter 3, authors introduced a multi-UAV system for disaster response system. Tracking mission and scanning mission used image processing techniques in order to make an evacuation map from captured images of pedestrians and ar-eas, it also requires high computational processing capabilities to distinguish valuable data and control UAVs at the same time. So, in this Chapter 4, author is proceeding the work with intensely focusing on enhancement of the multi-UAV communication control and network, suggesting using swarm clouds as multiple GCS nodes instead of a single cloud. Author has inspired by swarm clouds after it has been verified in work of [Ardagna, 2012].

4.2.1 UAV as a device

Utilizing UAV as an IoT device, UAV must possess a capability to connect to the Internet.

Hence, the platform of the UAV must be compatible with cellular, and wireless communi-cations. Nowadays, the technology of embedded systems offers comprehensive features for engineers to configure, develop and integrate devices mutually together. In this study, au-thor used Linux single board comupter on the main frame of the UAV. It helps integrating UAV board into a cellular dongle. Moreover, author used virtual proxy network (VPN) ser-vices to earn a secure connection and identify GCSs and UAVs IPs. Figure 4.1 demonstrates the proposed system of single UAV connected to the cloud via a VPN service.

4.2.2 System design

Any simplistic IoT architecture consists of devices, networks, and applications domain (see Figure 4.1). In Figure 4.2, Multi-UAV controlled by one GCS Cloud. Technically, GCS-cloud manages UAVs and transmits and receives data based on UAV System Identification (SYS ID), which is a systemic identification for each UAV in the network. In this configuration, MAVProxy is the master protocol to control UAVs [Nirmala, 2014]. Also, specific ports have been assigned to provide a simultaneous and real-time control system.

4.2.3 Swarm clouds as Multi-GCS

Swarm clouds are more efficient to control multiple UAVs since each cloud can be assigned to a single UAV or a group of UAVs depending on the system design and cloud processing ability. Also, a third-party of clouds can analyze data, monitor the operations, and record the flight history like logs for each UAV. PPTP protocol has been used to have secure communi-cation between the VPN server and UAVs. The developed Multi-GCS is shown in Figure 4.3.

Both of Figures 4.2 and 4.3 demonstrated independent flights; however, shadow control can

VPN Server

GCS Cloud

Cellular Connection

Internet Internet

Internet

Remote Client

UAV Access Point

Figure 4.1:UAV controlled by one cloud GCS via VPN.

Cloud Devices

PPTP

VPN Server UAV1

SYS_ID1

MAVlink

MAVlink

MAVlink

Interface and Command UAV2

SYS_ID2

UAVn SYS_ID3

Assignments and Tasks Networks

Data Processing Connection Authentication 10.8.0.1:14550 GCS

10.8.0.2:14550

10.8.0.X:14550

10.8.0.4:14550 HTTPS

Telemetry and Images

Data

PPTP TCP port 1723 PPTP TCP port 1723 PPTP TCP port 1723

Figure 4.2:Multi-UAV on one cloud GCS system.

deliver central monitoring and management, which makes UAVs completely visible to the client and accessible via remote protocols such as Remote Desk Protocol (RDP) and Secure Shell Protocol (SSH).

PPTP TCP port 1723

PPTP TCP port 1723

Swarm Clouds Devices

VPN Server MAVlink UAV2

SYS_ID2

Networks

GCS Control Interface

GCS Control Interface

GCS Control Interface Telemetry

Images Data Telemetry

Images Data Telemetry

Images Data

All logs and Data All logs and Data All logs and Data

Data Process

Shadow Control MAVlink

UAV1 SYS_ID1

MAVlink UAVn

SYS_IDn

HTTPS HTTPS

HTTPS

RDP Port 3389 SSH : 22

SSH : 22 SSH : 22 PPTP

PPTP

PPTP 10.8.0.1:14550

10.8.0.2:14550

10.8.0.X:14550

10.8.0.4:14550

10.8.0.5:14550

10.8.0.X:14550 PPTP

TCP port 1723

Third-Party

GCS IP GCS IP

GCS IP

Figure 4.3:Developed system designed for Multi-GCS system.

4.2.4 Implementation

This system demonstrates two UAVs controlled by swarm clouds with and assigned with different flight modes. ArduPilot Mega (APM) flight modes is used to control UAVs [Bin, 2009]. The guided initial UAV to receive all waypoints from the cloud server. Then, Auto mode fetches all received waypoints to start autopilot mission. "Stabilized" mode is used for manual assistance and shadow control. It also uses onboard sensors to perform a sta-ble flying experience. In both of "Loiter" and Auto modes system employs 3D-GPS lock to perform better flight experience. Failsafe and Return to Launch (RTL) have been stipulated for emergency (i.e., low battery or lost connection). Figures 4.3, 4.4, and 4.5 illustrate two UAVs on different modes of control by swarm cloud after executing scanning mission in real experiments at the institution campus.

4.2.5 Summary

This system shows an invented method to control multi-UAV throughout swarm clouds servers. Each cloud server has been assigned to control one or a group of UAVs, and third-party clouds can manage the income data from the flight controller to manage UAV’s assign-ments. In this system, the performance of multi-GCS proved its dependability compared to a single-GCS system, communication was more consistent due using multiple network ports, and the approachability to the UAV controller was much faster considering each UAV has a sole connection to individual private cloud. The system has not experienced any overload on swarm clouds as he has on a single cloud system, which is reasonably anticipated. How-ever, communication requires more evaluation. Therefore, the next section will go in-depth among TCP and UDP protocols evaluation. Furthermore, it describes how UAV can be a valuable device if it converts to an IoT device.

Figure 4.4: Navigation data of two UAVs on different GCS-Cloud on real flight experiment.

Path of UAV 1 Path of UAV 2 120

100 80 60 40 20 0 -20

12:07:00 12:09:00 12:11:00 12:13:00 12:15:00 12:17:00 Time

Altitude (m) UAV1 UAV2

Start of Guided Mode

Start Points of Auto Mode

Start of RTL Mode

Stabilized Mode

Figure 4.5:Graph of altitude and time for two UAVs in mission and third party cloud represents logs of each UAV.