4.6.6. Tampering

Sensors can be placed either indoors or outdoors. Indoor sensors can be easily managed and protected, while outdoor sensors are more vulnerable to attackers due to remote locations with poor security, harsh climates, etc 78. The probability that these sensors will be physically attacked is much higher; therefore, physical protection cannot be guaranteed. A DoS attack manipulates the network by breaking the connection or changing the current network. The attacker can also replace the original node with a fake or malicious node, causing a severe attack on the network [92]. In a Sybil attack, a malicious node penetrates each cluster head of the network and affects the operation of the routing protocol. Compromised nodes can be used to launch new attacks without exposing themselves [41,93]. These nodes are difficult to detect and isolate, allowing an attacker to alter data or transmit malware throughout the network that causes significant damage [94]. Constant monitoring of the network is necessary to ensure that WSN nodes cannot be tampered with and that network performance remains stable [95].

### 4.6.7. Authorization and Authentication

Nodes are the building blocks of the Internet of Things that must be defined in the network. Transmission between devices and access to the entire network span a wide range in IoT and WSN. IoT devices perform role-based access control, and their devices are allowed to do only what is required [96,97]. Devices and their data must be protected from physical and logical attacks on the network. Attacks on the DNS cache could affect the overall performance of the network. Authentication is the process by which each node on the network can access data based on a fixed connection to a server or cloud-based server. If the authentication process is not administered properly, it will lead to security issues and questions. An attacker can easily access the network and make it fail temporarily by making too many wrong attempts.

Authentication is complicated due to the massive proliferation of wireless media and the nature of sensor networks. Authentication is usually done using the credentials of a legitimate user [98,99]. However, this technique is not secure enough. Therefore, passwords should be changed regularly and computers should not be left unattended to make this technique robust. Both the sender and the recipient should perform authentication to verify the origin of the communication [100,101].

### *4.7. RQ7: Limitations of the Literature Review*

In this section, Table 9 explains the proposed solutions of the work conducted by various authors and the contributions with the limitations of their work are also described. The goal was to find research gaps in this area to help other researchers. The research gaps will allow researchers to develop solutions and new methods that could help fill the missing piece.


**Table 9.** Contributions and limitation of the literature.


**Table9.** *Cont.*




### **5. Challenges and Open Issues**

Intelligent systems can address various problems faced by industry, but there have been some challenges in integrating IoT and WSN into Industry 4.0. Technological improvements in IoT and WSN have increased concerns about security and data managemen<sup>t</sup> [96]. As more and more data is generated, it is difficult for factories and industries to manage it properly. Artificial intelligence algorithms have been implemented to manage Big Data and make systems and devices act more intelligently. The algorithms are used to process the data in different time periods. For education, the data must be shared in a central repository, while enterprises are mainly reluctant to share their private data due to poor

and insufficient organizational support for data in Industry 4.0. There are also safety managemen<sup>t</sup> issues in Industry 4.0 [116].

**Big data**: The emergence of various technologies and the explosion of their use have led to the outstanding development of Big Data technology and processing. Every device produces a huge amount of data. Due to the growing amount of big data, the improvements of Big Data packages encounter limitations and demand situations that need to be "overcome" in order to manage the amount of data used efficiently.

**Adapting to 6G**: 6G is another challenge for wireless sensor networks and the Internet of Things. Processing power is a major challenge in developing low-power and low-cost 6G devices. In addition, 6G brings privacy and security challenges for WSN and IoT.

**Updates**:system components could not be upgraded due to interoperability between protocols, systems, and their components. Therefore, systems are more vulnerable to attack if any part of a single system from a network is infected in intelligent factories.

**Environment**: security is also a critical challenge in WSN [97,100,106]. WSN nodes are not secure when deployed in a prone environment due to the wireless transmission of data. An attacker can access them from anywhere in the world and manipulate them easily. Internet attacks can also affect the vulnerability of sensor nodes.

**Supply chain management systems**: IoT devices are spreading erery day, posing new challenges to the integrity and scalability of supply chain managemen<sup>t</sup> systems [117,118]. Simultaneously connecting IoT devices to the cloud or the Internet requires a lot of access control, fault tolerance, data management, privacy, and security.

**Limited resources**: are other challenges in WSN domain that affect the energy of sensor nodes used in the network. Sensor nodes usually change their mode from sleep to active and vice versa. Therefore, sleep mode is considered as "outside the network" while active mode brings some other issues such as energy consumption [119]. Due to the high energy consumption, they also became dead. Sensor nodes usually have limited power, processing, and memory. In addition, sensor mobility is another problem that hinders the integration of mobile sensor nodes with the Internet.

### **6. Future Directions**

Industry 4.0 leads to the merging of people and technology to complement human activities with intelligent machines. Industry 4.0 will lead to customized human fashion that will minimize the oversupply and unavailability of supplies or items. Human-machine interaction will increase productivity and customer satisfaction with customized products.

The next version of Industry 4.0 is Industry 5.0, which is expected to be more userfriendly and better integrate technology with society and the environment. It depends mainly on robots. Robots are already the backbone of manufacturing, and Industry 4.0 technologies [120,121] provide flexibility in manufacturing. Industry 5.0 combines human creativity and craftsmanship with the speed, productivity (e.g., CPS) [122], and consistency of robots. In this version, robots can be programmed to work alongside humans.

**Soft computing**: can be used to reduce the dimensions of these large dimensional data sets [123]. Good features are essential to make efficient decisions. This is the reason why soft computing techniques are used to obtain useful features.

**Explainable artificial intelligence (XAI)**: can be used for interpretability of the decision made by the classification model. Classification models make decisions in a black box where the user does not know how the decision is made. XAI converts this black box into a white box and interprets the decision made by a model. XAI increases user confidence to take further action [124].

**Federated learning (FL):** is an optimal choice for privacy preservation. FL works by training global and local models on the edge device. The model on the edge device does not share the data with the global, thus keeping the data private at each edge device. Only parameters are shared globally to retrain the global model and optimize the inference results [125,126].

**Secure devices**: sensor nodes are designed to consume less energy and become active when they are needed or an event occurs [59]. Further improvements are also needed to prevent attacks from the Internet. While the IoT has no limitations in terms of processing or energy. Due to the tremendous proliferation of IoT devices, this paradigm is now being shifted from the IoT to the Internet of Everything.

**Sustainability**: IoT systems are now moving toward the idea of self-organization, and systems are becoming capable of responding in an automated and adaptive manner and dealing with changes and uncertainties in the environment [118].

**Education 5.0**: in this digital era, education must also change from traditional to integrating hardware and software with co-bots to develop new skills and a smart society. Educational institutes are now using pedagogical tools to provide a better experience. Even though IoT-based education is still not widespread, there is still room for further improvement, such as sensor node coverage and efficiency, wireless data transmission of data [127], battery life, and high-cost nodes.

**General directions:** there are many challenges. Future directions may address heterogeneous interoperability of systems, self-organization protocols, routing schemes for managing IoT networks, data managemen<sup>t</sup> [79], cross-platform optimization, and the development of network security algorithms to secure wireless transmission from data manipulation, stealing, or hacking activities. On the hardware side, researchers are developing energy-efficient sensor nodes [91,115], with net-zero power to reduce maximum power consumption.
