Ubiquitous Control of a CNC Machine: Proof of Concept for Industrial IoT Applications

: In this paper, an integrated system to control and manage a state-of-the-art industrial computer numerical control (CNC) machine (Studer S33) using a commercially available tablet (Samsung Galaxy Tablet S2) is presented as a proof of concept (PoC) for the ubiquitous control of industrial machines. As a PoC, the proposed system provides useful insights to support the further development of full-ﬂedged systems for Industrial Internet of Things (IIoT) applications. The proposed system allows for the quasi-decentralisation of the control architecture of conventional programmable logic controller (PLC)-based industrial control systems (ICSs) through data and information exchange over the transmission control protocol and the internet protocol (TCP/IP) suite using multiple agents. Based on the TCP/IP suite, a network device (Samsung Galaxy Tablet S2) and a process ﬁeld net (PROFINET) device (Siemens Simatic S7-1200) are interfaced using a single-board computer (Raspberry Pi 4). An override system mainly comprising emergency stop and acknowledge buttons is also conﬁgured using the single-board computer. The input signals from the override system are transmitted to the PROFINET device (i.e., the industrial control unit (ICU)) over TCP/IP. A fully functional working prototype is realised as a PoC for an integrated system designated for the wireless and ubiquitous control of the CNC machine. The working prototype as an entity mainly comprises a mobile (handheld) touch-sensitive human-machine interface (HMI), a shielded single-board computer, and an override system, all ﬁtted into a compact case with physical dimensions of 300 mm by 180 mm by 175 mm. To avert potential cyber attacks or threats to a reasonable extent and to guarantee the security of the PoC, a multi-factor authentication (MFA) including an administrative password and an IP address is implemented to control the access to the web-based ubiquitous HMI proffered by the PoC.


Introduction
Due to the enhanced global interconnection of massive computer networks, the internet is now accessible to about 70% of the world's population [1].In the years to come, this percentage is expected to increase because giant technological companies such as Amazon, SpaceX, and Virgin Galactic are investing heavily in research and development efforts that will ultimately guarantee ubiquitous access to the internet [2,3].Today, typical end-user devices such as mobile devices (e.g., phones, tablets, and others), personal computers, and others constitute part of larger networks whose interconnections have been made possible by the internet [4].This global infrastructure, which primarily enables the connection of all kinds of devices to allow for augmented or enhanced wireless communications and the sharing of information between devices, is generally referred to as the Internet of Things (IoT) [5].
Apart from the robust wireless connectivity of devices, the IoT also allows for remote or virtual human-to-human, human-to-machine, and machine-to-machine interactions [6][7][8].These and many other capabilities (such as pervasiveness or ubiquity) of the IoT are what makes it attractive for many specialised applications such as process monitoring and control across many industries, which has led to the advent of the Industrial IoT (IIoT) [5,9].While IIoT and its applications in and of themselves are not posited to usurp or replace existing industrial systems such as legacy automation and manufacturing systems [10], IIoT paradigms are bound to offer expedited means of ensuring smart monitoring and intelligent control of these systems for improved productivity and reliability [11].A case in point attracting a great deal of interest is the full decentralisation or quasi-decentralisation of the control architectures of traditional industrial control systems (ICSs) enabled by multiple agents that are connected by the internet and act alongside industrial control units (ICUs) to support the ubiquitous control and/or monitoring of industrial systems [12][13][14].
To develop a human-machine interface (HMI) for the ubiquitous control of industrial machines and equipment as a paradigm for IIoT applications, the control units or systems for industrial machines such as computer numerical control (CNC) machines must firstly have a dedicated connection to a standard industrial data network such as process field net (PROFINET) [15].Because programmable logic controllers (PLCs) (the dedicated control units for most industrial machines and equipment) are designed to have network ports that are compatible with the transmission control protocol and internet protocol (TCP/IP) suite, it is possible to make a PLC-based industrial machine or equipment a node on a larger network such as the internet.To implement this, the primary challenge is to have a device that routes signals between the machine's PROFINET (localised industrial network of the machine without internet connection) and the Ethernet (local area network for internet connectivity).Additionally, purpose-built technologies such as single-board multipurpose computers with on-board micro-controller units (e.g., Raspberry Pi [16]) will be required for the transmission of digital signals from the HMI (now made ubiquitous over TCP/IP) to the control unit of the machine over TCP/IP.This sort of control architecture poses another challenge in terms of cybersecurity.This is because the industrial machine and its peripherals go online over TCP/IP, making the industrial system in effect a cyber-physical system (CPS) [17,18].Furthermore, since the ubiquitous HMI is expected to be portable and mobile, a handheld device with a very low profile in terms of size and weight and an override system (i.e., emergency buttons) that can be easily accessible by system operators' fingers to ensure speedy interventions are required.
In this work, a handheld lightweight integrated system is proposed as a proof of concept (PoC) for the ubiquitous control of industrial machines to overcome the challenges and meet the functional requirements described above.The proposed PoC offers the following features which are in accordance with state-of-the-art paradigms for the remote, quasi-decentralised and pervasive control of industrial machines and equipment in line with present-day IIoT trends:

•
Display of the states of the industrial machine via a web-based ubiquitous HMI, making the industrial system in effect a CPS; • Multi-factor authentication including an administrative password and the IP address of the industrial machine to control access to the web-based ubiquitous HMI of the proposed PoC, to avert potential cyber attacks or threats that are often associated with wireless communication systems such as IIoT systems; • Touchscreen-based input configurations to manage control actions on the industrial machine over TCP/IP for the remote, quasi-decentralised and ubiquitous control of the industrial machine through the use of multiple agents (a Samsung Galaxy Tablet S2 and a Raspberry Pi 4), acting alongside the traditional PLC-based ICU; • Conventional override system achieved with emergency and acknowledge actions on the industrial machine via digital signals from push buttons implemented over TCP/IP for the more robust, remote, quasi-decentralised, and ubiquitous control of the industrial machine through the use of multiple agents (a Samsung Galaxy Tablet S2 and a Raspberry Pi 4), acting alongside the traditional PLC-based ICU.
The remainder of the paper is organised as follows: Section 2 summarily discusses the related work to reemphasise the practical needs of the work carried out; Section 3 describes the adopted devices and systems for the implementation of the PoC; Section 4 details the modi operandi of the working prototype of the PoC, including the configurations and operations of the adopted devices, and a performance evaluation of the PoC; Section 5 highlights the challenges and limitations of the working prototype of the PoC; and the concluding remarks are provided in Section 6.

Related Work
Due to new intelligent and innovative operations such as remote configuration, selfdiagnosis, etc., present-day industrial production and manufacturing processes demand highly robust operations and the quasi or full decentralisation of the control architectures for the integrated and distributed control of the machines and equipment deployed on the shop floor [12,19,20].The extent of decentralisation will always be relative to the architecture or arrangement of the ICSs and the coupling between the subsystems that make up the ICSs.Robust operations (e.g., pervasive sensing for time-critical applications [21]), guaranteed functional safety of electrical drives and machines (e.g., forced shutdown or emergency stop of industrial machines in line with global standards and practice [22]), and improved energy efficiency (e.g., energy-efficient or optimum pump operations in cooling systems [23]) can often be realised on ICSs through the use of mobile or ubiquitous HMIs acting as multiple agents or add-ons or complementary platforms alongside traditional ICUs (typically, PLCs) [7,24].Depending on their arrangement and operations, these ubiquitous HMIs allow for the quasi or full decentralisation of the control architectures of traditional ICSs.
The quasi or full decentralisation of the control architectures of traditional ICs is required because the conventional centralised control architecture, which overly relies on a single agent or component (usually, a PLC) to be the designated controller, does not always support the management of information structure constraints, uncertainties, delays, and a large span of control in the physical absence of a system operator on the shop floor [20].Present-day high dimensional large-scale complex industrial systems often require robust remote operations in the absence of a system operator on the shop floor [25,26].As a result, fully decentralised and quasi-decentralised control architectures offer the advantages of ease of expansion of the ICS, better localisation of faults, better supervision and control, and more robustness in comparison to conventional fully centralised control architectures [13,14,[25][26][27].
In [24], to make industrial process visualisation and supervisory plant process monitoring process more decentralised, flexible and robust, non-conventional HMIs (power-wall and table) are proposed and investigated using an experimental agent-based HMI for supervisory control and data acquisition (SCADA).In principle, the paradigms in [24] can conveniently be adopted to design and develop ubiquitous or online SCADA systems for industrial applications.However, to ensure the mobility of the remote control architecture, the physical dimensions of the HMIs and the overall remote control architecture in [24] could be challenging to implement on a small scale with a low profile.
To demonstrate the feasibility of employing mobile HMIs for the decentralisation and ubiquity of ICSs, a framework that allows for the open-source design and development of native mobile Android-based HMIs for ICSs is proposed in [7].The methodology is demonstrated via successful home automation, sewage treatment plant, and traffic lights control system case studies.Even though these case studies circumvent the cost overhead associated with proprietary solutions [28] (e.g., MobileHMI [29]), actual sensors have not been connected to the PLCs in the implementation as expected in real practice.Additionally, the implementation also requires a number of ad-hoc processes such as the generation, verification and execution of the Android code.
To support IIoT paradigms that enable the digitisation of the shop floor and subsequent decentralisation of the control architectures for ICSs, today, many leading manufacturers of industrial automation and control products offer products that provide the flexibility of innovative user-end design and development of mobile HMIs on a number of mobile operating systems [30,31].Siemens AG tends to be at the forefront of this, with awardwinning products such as their range of Simatic PLCs [32,33].Specifically, in [34], an IIoT paradigm for the control of industrial machines is presented in which the Simatic HMI comfort panel designated for operating industrial machines can be accessed remotely by a standard computer through a virtual private network (VPN).
Such a VPN can be leveraged to implement low-cost, low-profile mobile HMIs by simply making the VPN accessible on compatible handheld or mobile devices over TCP/IP.Even though such mobile or ubiquitous HMIs will adequately support the quasi or full decentralisation of the control architectures of their associated ICs, as discussed above, they also make their ICSs prone to cyber attacks or threats due to their dedicated internet connectivity [35].Several frameworks and/or methodologies such as the National Institute of Standards and Technology (NIST) framework [36,37] are available today to assess cyber attacks or threats such as denial of service [38] and devise the means of mitigating them in industrial applications, but cybersecurity has remained an area of ongoing research due to its very broad and dynamic nature in industrial applications [39].
In this work, a low-cost and low profile quasi-decentralisation of the conventional PLC-based control architecture for a state-of-the-art CNC machine (Studer S33) is achieved through data and information exchange between multiple agents (a Samsung Galaxy Tablet S2 and a Raspberry Pi 4) and the dedicated ICU (Siemens Simatic S7-1200) of the CNC machine over TCP/IP.The quasi-decentralisation achieved adequately supported the ubiquitous control of the CNC machine through the configuration of the first agent (Raspberry Pi 4) as a dedicated wireless access point (WAP) on the control architecture of the investigated ICS.In this way, the additional agent (Samsung Galaxy Tablet S2) is made an actor within the control architecture of the investigated ICS via a web-based application.Consequently, the quasi-decentralisation made the investigated ICS in effect a CPS.Since the proposed PoC causes the ICS to double as a CPS, the primary gateways for potential cyber attacks or threats are identified to be the WAP configured on the first agent (Raspberry Pi 4) and the web-based application configured on the second agent (Samsung Galaxy Tablet S2).
Following the recommendations in [40], a multi-factor authentication (administrative password and IP address) is adopted in this work to prevent the web-based application on the Samsung Galaxy Tablet S2 from unauthorised access and transferring sensitive information.To secure the Raspberry Pi 4 from potential cyber attacks or threats, the default username and password of its operating system (OS) were changed to a more secure username and password comprising alphanumeric and symbols as recommended in [41].In addition, the port of the secure shell protocol (SSH) on the OS of the Raspberry Pi 4, which allows for remote log on, was closed to ensure that it cannot be accessed remotely for reconfiguration as suggested in [41].
In comparison to some of the recent works discussed above, the proposed PoC is relatively secure, and it offers a higher degree of simplicity in terms of configurations and affordability (considering development and operational costs), while achieving a quasi-decentralisation of the control architecture of the investigated ICS to allow for the ubiquity of control operations.In the subsequent sections, the proposed PoC and its features are discussed to reveal how it might support the development of full-fledged systems for the full decentralisation of the control architectures of traditional ICSs in line with contemporary IIoT paradigms.

Adopted Devices and Systems
The layout of the proposed PoC is shown in Figure 1.In Figure 1, the essential devices and systems that constitute the proposed PoC are a tablet with HMI (Samsung Galaxy Tablet S2 [42]), a PROFINET slave/device (Raspberry Pi 4 [16,43]), an Ethernet switch (Netgear GS305E switch [44]), and a PROFINET master, a PLC (Siemens Simatic S7-1200 [45,46]) integrated with a state-of-the-art CNC universal cylindrical grinding machine (Studer S33 [47]).As shown in Figure 1, all the devices and systems are connected to each other via physical cables, except the tablet with HMI, which bidirectionally communicates with the PROFINET Slave/Device using a wireless internet link (i.e., over the TCP/IP).Note that the link between the PROFINET Slave (Raspberry Pi 4) and the Ethernet switch (Netgear 305E) could be implemented using a physical cable (an Ethernet cable) or a wireless medium (Wi-Fi access point).It is essential to have a wireless connection between the PROFINET Slave and the Ethernet switch to demonstrate the full extent of the ubiquity of the control mechanism the proposed PoC offers.In practice, the tablet with HMI is to be collocated with the PROFINET slave to support physical connections to external compatible modules for sensing and actuation .The devices and systems that constitute the proposed PoC are summarily described as follows to highlight their specific functions.

CNC Machine
The industrial machine to be controlled in the proposed PoC is a CNC universal cylindrical grinding machine (Studer S33) [47].The Studer S33 is a state-of-the-art industrial CNC machine designed to cut different sizes of medium workpieces at a cutting speed of up to 50 ms −1 .It can be equipped with different grinding tools, and it supports both individual and batch production of medium-sized workpieces.Using purpose-built application software (StuderPictogramming and StuderWINgramming), the grinding process on the Studer S33 can be configured or programmed either using the HMI of its relatively portable control unit (Fanuc 0i-TF [48]) or a personal computer (PC) linked to its control unit.A constant cutting speed can be configured for the Studer S33 using a turret wheelhead with a frequency-controlled motor spindle.More technical details about the S33 can be found in [47].

PROFINET Master
The PROFINET master in the proposed PoC is the Siemens Simatic S7-1200 [45,46], as shown in Figure 1.It is a device that provides the required degree of flexibility and power to control a wide variety of industrial machines such as CNC machines.It has a compact design, robust instruction set, and flexible configuration to support many industrial applications.Its central processing unit (CPU) comprises a microprocessor, an integrated power supply, input and output circuits (I/Os), built-in PROFINET, high-speed motion control I/O, and on-board analog inputs.Due to its flexibility and robustness, the Siemens Simatic S7-1200 can effectively and efficiently control large machines such as robot manipulators using complex and dynamic programming cycles.Additionally, its I/Os can all be employed at the same time by adding additional units, in which each has about 20 inputs and 20 outputs.The central processing unit (CPU) of the Siemens Simatic S7-1200 provides a PROFINET port for communication over a PROFINET network.Additional modules are also available for communicating over the process field bus (PROFIBUS), RS485 and RS232 networks, and Ethernet networks.In this way, the Siemens Simatic S7-1200 can link PROFINET and Ethernet devices together and be programmed to send specific commands to an IP address such that nodes on the network can control and/or read the I/Os.More technical details about the Siemens Simatic S7-1200 can be found in [45,46].

Ethernet Switch
To link the devices and systems in the proposed PoC, as shown in Figure 1, Ethernet cables with RJ45 connectors and an Ethernet switch are required.The Ethernet switch adopted in the proposed PoC is the Netgear 305E switch [44].Note that other Ethernet switches (including Ethernet switches that have an integrated Wi-Fi (wireless fidelity) access point) can be used.The Netgear GS305E switch has a buffer size of 128 kB, a performance bandwidth of 10 Gps, and a meantime between failure of 2,531,294 h at 25 • C. It operates using the IEEE 802.1p class of service (COS) and a port-based virtual local area network (VLAN).With its integrated VLAN function, the Netgear GS305E switch supports the configuration and establishment of different networks.More technical details about the Netgear GS305E switch can be found in [44].

PROFINET Slave
The PROFINET slave or device in the proposed PoC is the Raspberry Pi 4, as shown in Figure 1.The Raspberry Pi 4 is basically is a single-board computer using a Linux operating system.The Raspberry Pi 4 can be deployed as a computer with a graphical user interface (GUI).Even though the Raspberry Pi 4 is not as powerful and robust as the better-known personal computers with more computational power and capabilities, it is capable of carrying out several computational and processing tasks.It comes with an Ethernet port and a Wi-Fi module to allow for network connectivity, and it has digital I/Os that are compatible with analogue-to-digital converter chips.In this way, the Raspberry Pi 4 is capable of working with analogue data or signals.Due to this functionality, push buttons and industrial machines such as motors can be connected to the Raspberry Pi [16].For wireless connectivity, the Raspberry Pi 4 uses a dual-band IEEE 802.11b/g/n/ac wireless that can run at 2.4 GHz (licensed industrial, scientific and medical (ISM) band) or 5.0 GHz (Unlicensed National Information Infrastructure (U-NII) band) [49].More technical details about the Raspberry Pi 4 can be found in [43,49].

Tablet with HMI
The proposed PoC uses the Samsung Galaxy Tablet S2 as its tablet with HMI, as shown in Figure 1.The Samsung Galaxy Tablet S2 has a multi-touch touchscreen and an on-board Wi-Fi module that supports the IEEE 802.11 a/b/g/n/ac wireless networking standards, allowing for dual-band connectivity, 2.4 GHz or 5 GHz, similar to the Raspberry Pi.Since several Android OS-based applications such as internet browsers can also be installed, configured, and deployed on the Samsung Galaxy Tablet S2, it is highly compatible for the deployment of HMIs [50].These features are harnessed to have a web-based HMI on the Samsung Galaxy Tablet S2 for the proposed PoC as shown Figure 2.More technical details about the Samsung Galaxy Tablet S2 can be found in [42].

The Proposed PoC
The primary function of the CNC universal cylindrical grinding machine (Studer S33) used for the PoC is to grind metallic pieces mounted to a spindle.The spindle and a collaborative robot arm operating the grinding tool are controlled by a PLC (Siemens Simatic S7-1200 PLC in this case) and programmed to allow the movement and rotation of workpieces in different angles.In this way, the collaborative robot arm can drill at different velocity rates.When used cooperatively with the rotating spindle, the Studer S33 can be configured to grind various workpieces via an external control panel.The Samsung Galaxy Tablet S2 (Tablet with HMI) is configured to be the "control panel" for the PLC using PROFINET.Hence, Siemens Simatic S7-1200 PLC is the PROFINET master as described in Section 3.2.To access the Siemens Simatic S7-1200 PLC over PROFINET using the Samsung Galaxy Tablet S2, the PROFINET master is set up with a fixed IP address.By activating the web-server function of the PROFINET master and creating a website in HTML (HyperText Markup Language), a web-based HMI for the PROFINET master becomes accessible over TCP/IP from any device with a browser that fulfills the MFA (IP address and administrative password) as shown in Figure 3.In this way, connected devices automatically become part of the industrial network after fulfililng the MFA.
The web-based HMI is configured to send commands to the main programming logic that initiates the control operation sequences on the Studer S33.The conventional HMI for the CNC universal cylindrical grinding machine (S33) is the control unit (Fanuc 0i-TF [48]) as mentioned in Section 3.1.It is attached over a movable arm and has a size of about 1000 mm by 60 mm.These features indicate that the conventional panel is not very compact, and it is fixed at the location of the CNC machine.To program and configure the control unit of the CNC, a personal computer (PC) is first connected to the PROFINET master.The Siemens Totally Integrated Automation (TIA) Portal is used to both configure and program the PROFINET master.Since the TIA Portal is not designed to directly control the CNC machine, attempting to do so seems impractical; hence, in the proposed PoC, it is the PROFINET master (Siemens Simatic S7-1200) that is linked to the network used for the ubiquitous control of the Studer S33 and not the PC used to configure and program it.

Configuration
As mentioned in Section 3, a Samsung Galaxy Tablet S2 is linked to the Siemens Simatic S7-1200 PLC over PROFINET.This is done by configuring a Raspberry Pi 4 as a router to function as a dynamic host configuration protocol (DHCP) server and wireless access point (WAP) using a dual-band (2.4 GHz (ISM band) or 5.0 GHz (U-NII band)) Wi-Fi.By connecting both devices to the Raspberry Pi 4, a TCP/IP connection is established.The PROFINET master is configured according to Listing 1 and the Raspberry Pi 4 is configured according to Listing 2.

Begin
Open Siemens TIA portal Create a new project to configure Siemens Simatic S7-1200 Select device type = Hardware type Open the "General Settings" In "PROFINET Interface" Set the "IP Address" and "Subnet Mask" to according to the network Under "Advanced Options" Activate "Web Server Access" Under "Protection and Security" Activate the PUT/GET Option In "Pulse Generators (PTO/PWM)" Set "Startup after POWER ON" to "Warm Restart-RUN" In "Web Server" Create a new user and grant full access If (Device == Offline) Then Click "Go online" and "Start search" End If (Device == Found) Then Double click the device End End

Operations
The flow diagram of the operations of the proposed PoC is shown in Figure 3, and the communication system layout is shown in Figure 4.The essential steps are described summarily as follows: • Step 1: To control the CNC universal cylindrical grinding machine (Studer S33) ubiquitously, the system operator makes use of the touchscreen-operated Tablet (see Figure 3).By using a browser application, installed and configured on the tablet, the system operator enters the IP address of the CNC machine, and the web-based HMI created for the PROFINET master (Siemens Simatic S7-1200) can be accessed using an administrative password (see Figure 3).

•
Step 2: Following a successful authentication in Step 1, full administrative access is granted to the web-based HMI.Using the touchscreen of the tablet, the user can perform the following operations on the industrial motor: (1) ON ( 2) OFF ( 3) ACK (Acknowledge) (4) EMG (Emergency), as shown in the communication system layout in Figure 4.

•
Step 3: Depending on the system operator's selection in Step 2, while the ACK button is pressed down, the Studer S33 is "started" for ON or "stopped" for OFF.To "shut down/stop" the Studer S33 immediately, the EMG button shown in Figure 4 can used at any time.

•
Step 4: Once the Studer S33 is configured in Step 3, it remains in the configured state, and the entire process reiterates back to Step 1.

Begin
Configure the LAN connection of Raspberry Pi 4 Use a fixed IP-address in the file "/etc/dhcp.conf"In the same file (i.e., "/etc/dhcp.conf")Prioritize WLAN connection over LAN connection Activate the general-purpose I/Os (GPIOs) In the command line Implement "sudo apt-get update To request updates for Raspberry Pi 4 Implement "sudo apt-get install python-rpi.gpio" To insall python-rpi.gpioImplement "sudo adduser $USER gpio" To add user ("USER") to the group for the GPIOs In the command line Implement "sudo apt install hostapd" To installaccess point software package At the startup (booting) of the Raspberry Pi 4 Implement "sudo systemctl unmask hostapd" And Implement "sudo systemctl enable hostapd" To enable and start access point Implement "sudo apt install dnsmasq" To install the package for domain name system (DNS) and DHCP DHCP -distribute IP addresses to the connected devices DNS service -translate domain name of devices to IP addresses Implement "sudo DEBIAN_FRONTEND = noninteractive apt . . .install -y netfilter-persistent iptables-persistent" And Implement "sudo iptables -t nat -A POSTROUTING . . .

-o eth0 -j MASQUERADE"
To add firewall rules for data traffic sharinf between the devices Implement "sudo apt-get install netfilter-persistent -y" To install the service iptables-persistent to save firewall rules Implement "sudo netfilter-persistent save" To save firewall rules Add a static IP-address for "wlan0" in the file "/etc/dhcp.conf"In the command line Implement "sudo nano /etc/sysctl.d/routed-ap.conf"To enable the routing between ethernet and WiFi Add the text "dnsmasq.conf"to the newly created file To enable the routing by setting the configuration for the DHCP service Rename the original file In the command line Implement "sudo mv/etc/dnsmasq.conf/etc/dnsmasq.conf.orig"And Implement "sudo nano/etc/dnsmasq.conf"To have connected devices with IP addresses in the defined DHCP range Implement "sudo rfkill unblock wlan" To unblock WiFi Implement "sudo nano/etc/hostapd/hostapd.conf" To create a new file to configure Wi-Fi Define Wi-Fi service set identifier (SSID), password and others Reboot Device to connect and make access point go live Connect tablet to both Wi-Fi and Raspberry Pi 4 Access the website of the PROFINET master End

Analysis and Performance Evaluation
To analyze the operations of the proposed PoC and to evaluate its performance, its control operations must firstly be modelled mathematically to determine the best metric or set of metrics to adopt or employ.In this work, the control operations of proposed PoC can be modelled to address a minimum-time control problem [51], considering the state switching procedure of the Studer S33 (i.e., State 0: OFF ⇐⇒ State 1: ON, as shown in Figure 5) effected by the proposed PoC.Note that a full-fledged system should accommodate more states in the state switching procedure of the Studer S33; for example, the downtime, standby, idle, warm up, and operational states in the typical state switching procedure of a CNC machine tool [52].However, these states are not within the scope of our work due to the control operations configured on the proposed PoC.As a result, all states of the Studer S33 are concatenated into two main states (i.e., ON and OFF states) to reflect the primary operations offered by the proposed PoC (see Figure 5).It is envisaged that these two states are sufficient to support the investigative studies carried out in this work to guide the potential development of a full-fledged system that can take into account several states of the Studer S33.Using the two states of the Studer S33 that can be controlled by the proposed PoC, intuitively, it can be deduced that the control mechanism offered by the proposed PoC mirrors the conventional bang-bang controller that is often implemented for industrial machines that accept or can be configured to accept binary inputs such as the opening and closing of a switch [53].To model the bang-bang control signal (i.e., ON or OFF signal) from the proposed PoC, a formulation of the Heaviside step function can be used as follows: where t OFF and t ON are all the instances of time for which the Studer S33 is activated ON and OFF, respectively, by the proposed PoC, as shown in Figure 6.This follows from the assumption made that the Studer S33 remains in an OFF state for all instances of time until it receives an ON state signal from the proposed PoC or vice versa (see Figure 5).
To better understand the formulation in Equation ( 1), the latency (t L ) of the proposed PoC can be said to be directly proportional to the dead time (t D ) of its bang-bang control scheme.This is because the time taken for the proposed PoC to transfer data (i.e., ON or OFF activation signal) from the Samsung Galaxy Tablet S2 (i.e., tablet with HMI) to Studer S33-i.e., t L -will always be directly proportional to the delay from when the bang-bang control signal is issued until when the Studer S33 responds to the control signal; i.e., t D .This relationship can be expressed as follows: where k is the proportionality constant, which is assumed to be unity in our investigations for simplicity.This assumption is made because the individual time frames for t L and t D cannot be correctly ascertained to be distinguishable without rigorous analysis and investigations that are beyond the scope of this work.Hence, t L = t D is used in the subsequent investigations.For the typical operation of the Studer S33 using the proposed PoC (i.e., initialisation from t = 0, as the initial period), if the Studer S33 is initially in an OFF state (i.e., State = 0) based on the condition ∀ t = t OFF and it is expected to be in the ON state (i.e., State = 1) based on the condition ∀ t = t OFF , this cycle of operation can be modelled mathematically as shown in Figure 7.The plot of the mathematical model shown in Figure 7 can be described analytically as follows: If t OFF is ∀ t = t ON , by considering, t OFF , t ON and t L = t D , the Heaviside step function formulation for the bang-bang control scheme offered by the proposed PoC can be modelled mathematically, more accurately, as follows, for an ith operational cycle of the Studer S33: where t i D = t i L .From Equation (4) and Figure 7, it can be seen that t D or t L is a very important metric for the evaluation of the performance of the proposed PoC with respect to the operational cycles of the Studer S33.In this work, we have analysed the t D or t L of the proposed PoC over 100 operational cycles of the Studer S33 to ascertain the performance of the proposed PoC.The descriptive statistical analysis carried out is shown in Table 1 and the trends for the OFF and ON control states over the operational cycles are shown in Figure 8.
From Table 1, it can seen that over 100 operational testing cycles of of the CNC machine (Studer S33), the mean values of t D or t L for the OFF and ON activation states (i.e., 0.8040 s and 0.7919 s, respectively) are in close agreement and reasonably match the standard conventions for bang-bang control schemes designated for the control of physical systems [54].In addition, the worst values of t D or t L for the OFF and ON activation states (i.e., 0.9300 s and 0.9500 s, respectively) are still relatively fast.This shows that the proposed PoC is still efficient, even for the worst-case scenario.A low standard deviation of the t D or t L values for the OFF and ON activation states over the operational testing cycles-i.e., 0.0613 and 0.0656, respectively-indicates that the proposed PoC offers a very robust t D or t L .This is further corroborated by the trends in Figure 8.In Figure 8, it can be seen that t D or t L is relatively stable for the OFF and ON activation states over the operational testing cycles of the CNC machine (Studer S33).

Challenges and Opportunities
Using the tablet with HMI (i.e., web-based HMI), the CNC machine can be started or stopped.In addition, a maintenance mode can be turned ON or OFF, during which the acknowledge button (i.e., ACK in Step 3 in Section 4) needs to be pressed down for safety reasons.The emergency stop button (i.e., EMG in Step 3 in Section 4) is used to halt the running processes on the Studer S33 immediately in case of an emergency.To provide further control of the Studer S33, the web-based HMI can be programmed to manage hole grinding processes and display the status of running tasks and additional operational states of the CNC machine.Additional sensors and their readings can also be incorporated on the shop floor to provide status updates and other alerts, as depicted in Figure 4), to other connected devices (e.g., other portable devices) on the network.In this way, the connected devices can access and manage the Studer S33 and its operations on the shop floor from anywhere on the globe.In other words, the PoC effectively portrays a CPS-one of the core components of the IIoT.
Despite the ubiquitous control and robust digital footprints that the PoC affords the CNC machine (Studer S33), an "online" presence means that the machine can be accessed and controlled once the administrative and access control measures such as a firewall have been bypassed.As a standard configuration on the PROFINET master (Simatic S7-1200), MFA including an administrative password with the IP address of the Studer S33 is set to access the web-based HMI.In this way, even though the machine is "online", only authorised users can connect to the PLC's main control website and are able to access the web-based HMI based on the MFA, as shown in Figure 3.More factors such as key generation in the form of verification codes or one-time passwords (OTPs) sent to the operator via email or short message service can also be incorporated into the MFA as additional layers of access control to the web-based HMI to further alleviate potential cyber attacks or threats.
Due to the inherent constraints of web-based applications, programs for purpose-built web applications are often limited to compatible development languages.Hence, the level of interactivity of web-based HMIs could be limited as well because of the inherent constraints in programming constructs.Since the proposed PoC works over TCP/IP, the bandwidth and connection speed of the internet may also impose a t L that compounds the t D associated with controlling the CNC machine (Studer S33) via the web-based HMI.

Conclusions
To demonstrate the ubiquitous and wireless control of industrial machines, a PoC comprising a state-of-the-art CNC machine (Studer S33), a PLC (Siemens Simatic-1200), an Ethernet switch (Netgear 305E), a single-board computer (Raspberry Pi 4), and a commercially available tablet (Samsung Galaxy Tablet S2) is developed and investigated in this work as an IIoT paradigm.The ubiquitous and wireless control (primarily, ON and OFF operations) of the Studer S33 achieved via a web-based HMI on the Samsung Galaxy Tablet S2 validates the prospective use of the PoC in IIoT applications involving industrial machines on the shop floor.The profile, weight, and size of the working prototype of the mobile (handheld) touch-sensitive HMI, which fits into a 300 mm by 180 mm by 175 mm compact case, also demonstrates the enhanced mobility and ubiquity that the PoC offers in comparison to conventional or traditional industrial control units.In the future, the integration of sensors and sensing paradigms into the network to pervasively monitor processes and operations on the shop floor will be investigated using the PoC and findings in this work as a backbone for the development of a full-fledged system.

Figure 1 .
Figure 1.Layout of the proposed PoC showing the adopted devices and systems.

Figure 2 .
Figure 2. Tablet with HMI for controlling the CNC machine for the proposed PoC.

Figure 3 .
Figure 3. Flow diagram of the operations of the proposed PoC.

Figure 4 .
Figure 4. Communication system layout of the proposed PoC.

Figure 5 .
Figure 5. Concatenation of the state switching procedure of the CNC machine.

Figure 6 .
Figure 6.Mathematical model of the operations of the proposed PoC.

Figure 7 .
Figure 7. Mathematical model of the operations of proposed PoC for an operational cycle (ith cycle) of the CNC machine.

Figure 8 .
Figure 8. t D or t L trends for the OFF and ON states of the CNC machine over 100 operational testing cycles.

Table 1 .
Performance evaluation of the proposed PoC in terms of t D or t L in seconds over 100 operational testing cycles.