1 Introduction

Industry 4.0 has been attracting growing interest recently from researchers, governments, manufacturers, and application developers, since it can offer a reduction in energy consumption, increase economic benefits, and enable smart production. Industrial wireless networks (IWNs) are the key technology enabling the deployment of Industry 4.0. Most people are aware of the increasing advantages that wireless networks can offer, from experiencing easy access to high speed internet services using cellular phones, laptops, or other mobile devices. Wireless networks have a number of merits including flexibility, lack of wiring, and mobility, which has made mobile wireless networks popular in consumer electronics [18, 35, 51, 60, 81]. As industry and society have gradually developed, the concept of Industry 4.0 [62], smart factories [52], networking manufacturing [28] and other new frameworks have been proposed [82, 99]. Meanwhile, new information and communication technologies (ICTs), such as the industrial cloud [76], big data [19, 93], wireless cloud networks [75, 104], industrial internet of things [21] and high performance embedded system [74] have been introduced into the manufacturing industry to meet the demands for higher productivity, green production, higher market share and flexibility. In this regard, IWNs have been gradually entering industry vision, and are becoming vital foundations for realizing the architecture of Industry 4.0 and smart factories [31]. Although there are many surveys and reviews on IWSNs or IWNs, such as [35, 51, 60], few studies have considered the context of Industry 4.0 or smart factories. Furthermore, the new technologies arising from the great advancements of ICT must have an impact on IWN systems. Therefore, a review of IWNs in the context of Industry 4.0 is necessary, which is the main motivation for writing this paper.

As information has progressed in recent years, wireless communication technology and wireless sensor networks (WSNs) have been heavily researched and a large number of applications of WSNs in agriculture, military, health, and other domains have been applied [22]. In a WSN, the overall network contains a large number of ordinary nodes which each have a primary function for sensing physical parameters such as temperature, humidity, voice and other performances from the monitoring environment [12]. Generally, all WSN nodes are stationary, have limited power and do not consider the industrial environment and other special requirements, such as reliability, latency and flexibility. Furthermore, in the industrial domain, mobile nodes have been introduced into industrial systems incrementally [15, 27]. Wireless nodes or radio modules have been mounted on mobile devices to increase flexibility and mobility, which is also not considered by traditional WSNs.

IWNs have inherited many features from WSNs, especially communication protocols and industrial applications [91]. However, due to the particular features of the industrial environment and since Industry 4.0 industrial wireless networks are different from traditional WSNs [17, 37], there are some new constraints and requirements for IWNs. The main differences between IWNs and WSNs are listed as follows:

  • Latency IWNs are adopted in industry systems to sense important parameters such as monitoring the machine status and working environment or delivering control instructions and real-time information. Therefore low latency is required by these applications. In most cases, this is at the expense of energy consumption and a high cost to achieve real-time performance. In contrast, WSNs are constrained by energy and nodes may be deployed in unreachable domains making it difficult to replace or renew batteries. As a consequence, it is necessary to maximize battery life-time in WSNs resulting in higher latency.

  • Mobility As discussed above, to increase the flexibility and mobility for Industry 4.0, IWNs may contain more moving nodes, such as mobile products, mobile robots, automatic guided vehicles, unmanned aerial vehicles, workmen and other mobile devices. This is in contrast with WSN nodes, which are usually considered stationary or with few moving nodes such as a moving sink, relay nodes and so on.

  • Environments Another great difference between IWNs and WSNs is the operating environment. Firstly, in the industrial domain, IWNs operate in a challenging environment because of dust, vibration, heat, various obstacles, and a higher temperature and humidity. Secondly, there is more severe signal interference from motors and other wireless networks than for traditional WSNs. Furthermore, industrial environments can easily impact on the radio channel, which differs from WSNs. Therefore, IWNs need additional strategies to ensure reliability and efficient communication. In contrast, WSNs nodes are deployed in a relatively stable and friendly environment.

  • Capacity For Industry 4.0, all equipment, devices, workmen, terminals and other nodes can finish complicated tasks individually, and cooperate with other equipment. Especially for IWN, nodes not only communicate with their neighbors to complete mechanical tasks such as moving, clamping, and hauling, but also need to cope with problems such as signal interference, moving paths and data processing. As a consequence, from the perspective of node capacity, IWNs nodes require higher capacities to handle data processing, energy and storage, and are smarter than traditional WSNs nodes.

There are several aspects to our contributions as follows.

  • Firstly, a briefly layout of Industry 4.0, its framework, the characteristics of inter-factory IWNs and the function of wireless nodes are presented.

  • Secondly, a quality of service (QoS) and quality of data (QoD)-oriented architecture based on the features of Industry 4.0 and IWNs are discussed in detail, as well as existing solutions for QoS and QoD.

  • Thirdly, the current main standards and products are summarized and applications for IWNs are introduced. Finally, the new challenges and issues of IWNs are addressed.

This paper is structured as follows. In Sect. 2, we provide a brief introduction of Industry 4.0 and the characteristics of IWNs. We then describe and analyze the QoS and QoD architecture of IWNs from different aspects (real-time, reliability, longevity, security and privacy) in Sect. 3. Section 4 provides a survey and taxonomy of various applications for IWNs. Finally, Sect. 5 outlines the main challenges and research trends. Section 6 concludes this paper.

2 Context of IWNs

In this section, we discuss the context of IWNs, from both a general and a specific perspective, including Industry 4.0, IWNs and nodes. Firstly, a briefly description of Industry 4.0 is presented to illustrate the role of IWNs in this novel concept and framework. Then, an IWN schematic diagram is introduced to show the structure of general IWNs. We then review the possible roles of wireless nodes in IWNs from smart factory and Industry 4.0. Finally, based on this discussion, we summarize the characteristics required for IWNs to meet the new requirements of new concepts such as smart factory and Industry 4.0.

2.1 Industry 4.0

The objective of Industry 4.0 is to connect and integrate traditional industries, particularly manufacturing, to realize flexibility, adaptability, and efficiency and increase effective communication between producers and consumers [34, 77]. Industry 4.0 refers to cooperation between different factories that are generally located in different remote places. Therefore, communications and networks play an important role in Industry 4.0 [31].

Figure 1 illustrates the general layout of Industry 4.0. This framework contains four major components: a physical layer, a network, the cloud/big data and an application layer. Machines, robots, mobile devices, workmen, AGV and other intelligent entities constitute the physical layer for acquiring and computing data, completing mechanical tasks and other primary functions. The networks are formed by cellular, wired, IWNs, and other networks that transmit real-time data between different entities such as machine to machine (M2M), workmen to device, or networks to data servers or industrial clouds. The clouds (big data or servers) are responsible for data storage, cleaning, mining, high performance computing and other services [25]. Furthermore, the clouds provide the bridge between the network and application layers. The applications layer consists of the smart city, users, smart enterprises, and other smart services. It is anticipated that Industry 4.0 will be the next revolution for our society.

Fig. 1
figure 1

Layout of Industry 4.0

Full size image

2.2 IWNs

To further illustrate the framework of IWNs, we will now focus on wireless communication from the perspective of the plant or factory interior. IWNs actually have a similar structure in Industry 4.0, as shown in Fig. 2 (industrial wireless networks schematic diagram). The IWN communications system can be divided into four components: smart entities, inter-IWNs, beyond IWNs, displayers and servers. Within IWNs, smart entities (such as workmen, AGVs, machines, ordinary sensors) with wireless transceivers could be regarded as wireless nodes. It is clear that IWNs are formed between nodes by wireless radios. Beyond IWNs, the access point nodes and the gateway create a bridge to other networks such as cellular, wired, internet and other public networks [39]. Finally, higher-level data applications including data servers, management, controllers, and displayers, may be based on these specific networks.

Fig. 2
figure 2

Industrial wireless network schematic diagram

Full size image

For communication and networks within a factory, IWNs play a critical role in the whole system. Meanwhile, although industrial wired networks, e.g. the industrial internet, have made great progress using information and communication technologies, IWNs have become a good complimentary network to an industrial wired communication system [94], as shown in Figs. 1 and 2. IWNs have obvious advantages that coincide with the requirements of Industry 4.0, so more and more industrial wireless solutions have been adopted within factories, particularly where moving devices are used or wired networks are not suitable.

2.3 Wireless nodes

It is known that a smart industrial wireless node not only contains memory modules, a processor, a power source, RF transceivers, and sensors, but also mechanical parts and related control and communication programs [13, 45]. Moreover, in Industry 4.0, wireless nodes need to be smarter, more flexible and more mobile than traditional wireless network nodes. As a result, firstly, IWN nodes need to be mounted to stronger modules such as processors, memory, RF, energy and mechanical components. Secondly, wireless nodes should perform a variety of functions, such as relaying information, communication, and sensing parameters.

Compared with traditional wireless networks, industrial wireless nodes have higher capacities for processing, memory storage and energy. Meanwhile, with the development of IWNs, the scale of networks has become more and more complex, and as discussed above, wireless nodes need to complete many functions. To handle these growing requirements, nodes need to evolve from ordinary nodes into functional wireless nodes that can act with special roles. As explained in Fig. 2, IWNs can have various functional nodes because of different equipment and requirements. In Table 1, we have listed some of the major roles of wireless nodes in industrial environment from a communication and networking viewpoint.

Table 1 Roles of mobile node in industrial environment
Full size table

2.4 Characteristics of IWNs in Industry 4.0

IWNs have a high number of mobile nodes and many particular requirements for an industrial environment. The major important technical challenges in terms of their constraints, challenges and design goals for realizing IWNs 4.0 can be outlined as follows.

  1. 1.

    Dynamic topologies When mobile nodes are applied in industrial communication systems, it is obvious that the topology and link quality of the whole mobile WSN will dynamically change due to mobility, a hostile industrial environment, or node failures [11]. Additionally, at industrial sites in general, many protocols and networks may be operating at the same or neighboring frequencies (such as at 2.4 GHz). As a result, when a node is working in an industrial scenario, its routing path, clusters and neighboring nodes will be constantly varying as other nodes move and communicate. Therefore, dynamic topology is one of the key challenges for industrial IWNs applications [23].

  2. 2.

    Signal interference and path loss Generally in the industrial domain, mobile nodes move along the ground following a certain path, so their antennas have a relatively low height and thus these mobile nodes are susceptible to absorption of the wireless signal. Furthermore, mobile wireless nodes will inevitably face interference due to elements of the harsh industrial environment such as heavy dust, metal obstacles and other RF signals [32], which contravene wireless communication principles. Personal wireless mobile devices (such as mobile phones or laptops) may be present within the industrial field, and these devices will gradually aggravate signal interference. Meanwhile, received signals experience path loss and reflection interference based on the characteristics of wireless-wave propagation, resulting in wireless nodes with limited transmission distance [60].

  3. 3.

    Limited energy Many studies have illustrated that WSNs are still constrained by energy limits because wireless nodes are usually battery-powered [16]. However, in IWNs, batteries not only provide energy for wireless communication but also for mechanical systems, so the energy limitation is one of the challenges for prolonging network life [9]. Although, many WSN nodes use other energy sources such as solar or wind energy, and there have been many studies done to increase energy efficiency and save power [4], the energy limitation has not improved significantly in practical applications.

  4. 4.

    Changing location In IWN applications, especially in an industrial environment, the parameters of individual mobile nodes (such as their location, pathway, residual energy and so on) may be preconditions for completing other functions [1]. Moreover, it can be expected that the wireless network, monitoring and control center, and end-users may require detailed information on the location of some or all mobile nodes [90]. However, due to the mobility of wireless nodes in an IWN, localization becomes a difficult task [26]. Although many researchers have proposed a number of mobile node localization algorithms [65], there is still a lot of work required to overcome issues such as low accuracy, the impact of barriers, and lack of consideration of the industrial environment.

  5. 5.

    Physical collision and barriers When increasing numbers of mobile nodes are applied in the industrial domain, it is unavoidable that physical collision will happen between mobile nodes, equipment, workers and other barriers [20]. Since IWNs operate in the industrial scientific medical (ISM) band, wireless channels are a scarce resource [2]. Furthermore, the mobile wireless nodes need to communicate with neighboring nodes, cluster heads and the gateway, base-station or other nodes. However, in an IWN, there may only be one or a few carrier frequencies available. If there are multiple communications at the same time and in the same place, the data packets will collide [35, 58]. Additionally, there are many other objects such as equipment, workers, and moving bodies that may act as obstacles, having an unfavorable effect on the location and quality of communications [7].

3 Architecture of QoS and QoD-oriented IWNs

As we know, in the context of Industry 4.0, data (or information) and wireless communications is transmitted through IWNs. Big data, industrial clouds, and industrial systems are used in Industry 4.0 to increase productivity, reduce cost and energy consumption, and enhance flexibility and personalization. To realize the objectives of Industry 4.0, there are new requirements for the whole IWN system, especially for the service and the data. The data carries important information for the factory, services, manufacturing, and users and wireless communication is responsible for transmitting and receiving this data. Therefore, a new IWN architecture will be introduced, which aims to meet the demands for diversity and specificity of industrial applications. As shown in Fig. 3, QoS and QoD-oriented architecture of IWNs are configured for QoS and QoD of the entities. The entities are the foundation of the whole architecture, and include factory entities, networks facilities, and service equipment. The QoS indexes include real-time [85], reliability, longevity, security and privacy controls. The QoD performance covers validity, accuracy, reliability, and integrity. Furthermore, QoS and QoD are influenced by each other.

Fig. 3
figure 3

QoS and QoD architecture of IWNs

Full size image

3.1 Communication and networking

It can be concluded that IWNs play a primary role in the framework of Industry 4.0 and the smart factory. Generally, researchers use different layers to analyze the architecture of networks. IWNs are not an exception; many studies have divided the IWNs into a physical, medium access control (MAC), routing, transporting and application layer [47, 49, 51]. However, this traditional layer architecture cannot satisfy the increasing requirements for different applications which are arising from the continuous development of technologies. To address these challenges, the concept of a cross layer design has been proposed and successfully implemented [18, 72]. However, there are new QoS requirements that need to be satisfied for different IWNs in the industrial domain. As shown in Fig. 4, QoS of IWNs includes real-time, reliability, longevity, security and privacy criteria. To meet these QoS criteria, the design needs to be considered throughout all layers.

Fig. 4
figure 4

QoS of IWNs

Full size image

As IWNs have developed, more and more companies and alliances have invested in these domains. Many architectures and industrial protocols have been proposed, including Wi-Fi, ZigBee, Bluetooth, RFID, and other proprietary protocols. In Table 2, we have listed the main IWN communication protocols [18, 81]. Some comparisons of parameters such as the physical layer, standards, working frequency, and maximum throughput are listed. A complete list and more information can be found in [13].

Table 2 Comparisons of different communication technologies
Full size table

3.1.1 Real-time performance

WSNs have been applied in smart plants, industrial environment monitoring and automation factories for low latency wireless communication. Therefore, WSNs in industrial applications can be used to provide an example for IWNs. Real-time performance is an important index for the industrial QoS. Furthermore, communications between wireless nodes require low latency to improve productivity. However, different applications may have different real-time IWN requirements. Table 3 gives the latency requirements for different systems. Motion control systems have the fastest real time requirement (<10 ms), while monitoring systems tolerate the largest latency value (<100 ms).

Table 3 Real-time requirements in different industrial environments
Full size table

Many researchers have presented their methods to achieve good real-time performance from different views and layers of wireless networks. Most of these studies have achieved low latency by improving the routing, MAC, or transport layers or by using cross-layer mechanisms. The authors in [29] have introduced a new traffic scheduling algorithm that focuses on the industrial applications of WSN for real time services, based on a window scheduling algorithm and the IEEE 802.15.4 framework. The authors in [96] have presented a reliable real-time flooding-based routing protocol for IWSNs (REALFLOW) to achieve real-time performance, which generates a simple distribution method for the related nodes lists, rather than using traditional WSNs routing tables. The authors in [91] have proposed an SSA (segmented slot assignment) algorithm to enable real-time performance, which is based on TDMA and free nodes to enhance the real-time abilities of IWSNs by improving the efficiency of retransmissions to ensure that IWSNs will be adaptive to the dynamics of low power wireless links. In [69], an SAS-TDMA algorithm has been presented and simulated in TOSSIM to improve the QoS in time-constrained WSNs. Meanwhile, in [61] authors have introduced two-hop neighbor information-based gradient routing to enhance real-time performance while increasing energy efficiency. Each routing decision is based on the number of hops from source to sink and two-hop information. The authors have proposed a real-time message-scheduling algorithm for IEEE 802.15.4-based industrial WSNs. The scheduling algorithm can schedule a given periodic real-time message set, and the algorithm will determine the appropriate standard specific parameters to meet these timing constraints [94]. The authors in [55] have proposed a novel three-dimensional discrete-time Markov chain (DTMC) model for IEEE 802.11-based industrial wireless networks using the distributed coordination function (DCF) as the MAC mechanism. In [71], a novel real-time communication group sequential communication (GSC) scheme for 802.11e wireless networks has been introduced. This scheme improves the efficiency of the hybrid coordination function controlled channel access (HCCA) mechanism by reducing the protocol overheads of the 802.11e amendment. The study in [36] has proposed a (RT) protocol that simultaneously addresses congestion control and time-constrained event transport reliability objectives in wireless sensor and actor networks (WSANs). Table 4 gives a summary of different real-time algorithms.

Table 4 Real-time algorithms for IWSNs
Full size table

3.1.2 Reliability

In an industrial environment, the communication system is more susceptible to latency, particularly for automation, process control and manufacturing systems. Therefore, the reliability of the IWN is an important evaluation index. However, the moving wireless nodes and the harsh industrial environment of the IWNs can introduce more interference and increase the bit error rate (BER). In conventional WNSs, a retransmission mechanism is generally adopted to address these problems, which increases the latency. This is an issue for IWNs and therefore many authors have proposed efficient algorithms to guarantee network reliability [56]. In the following paragraphs, we will discuss the efficiency of approaches that are currently used to increase the reliability of WSNs in industrial applications.

Interference minimization As explained previously, IWNs have a high level of channel/packet errors due to multipath fading, the harsh industrial environment, and multiple existing wireless signals. To solve this problem and improve reliability, methods to minimize this interference have been developed. Specialized coding technologies can be another useful approach. Code division multiple access (CDMA) [41] is a classical technology used to minimize interference. However, with the expansion of network scales and industrial applications, CDMA has shown coding limitations in recent years. Another useful scheme is presented in [84], which assigns a pseudo noise (PN) code to each cluster in order to suppress inter-group interference via the code dimensions. In [44], the authors have proposed a deployment strategy to minimize the influence of interference, which analyzes, identifies and obtains the model steps using a statistical method. By analyzing the interference in WSNs with dual heterogeneous radios based on IEEE 802.15.4 and IEEE 802.11 standards, the author in [40] showed that using orthogonal channels for the two radios is not an effective solution to mitigate interference, and proposed an algorithm for adaptive packet aggregation and packet transmission scheduling.

Redundancy Since retransmission has many disadvantages for real-time performance, redundancy can be a better alternative option for industrial WSNs, especially for event-based WSNs. There are several ways to enable redundancy, including data redundancy and routing redundancy. The forward error correction (FEC) mechanism is a classical data redundancy method for improving reliability. In the FEC mechanism, the networks correct bit errors by adding redundant bits or data to the packet/frame [10]. Celimuge Wu et al. [83] have proposed a new protocol based on data redundancy that adaptively changes the redundancy level according to the application requirements and the link loss rate. The routing redundancy method uses the view that for the overall reliability of WSNs, any single transmission path is prone to weakness, especially in an industrial environment. To solve this problem, multiple routing/redundant routing algorithms have been proposed [67]. For example, Suraj et al. [68] presented a cluster based multipath routing protocol (CMRP) for improving the reliability and reducing the energy consumption of the WNSs, which computes the paths prior to usage. CMRP is a cluster-based routing protocol that requires a route from the cluster head to the base station. The base station provides the routing path of each sensor node in the network. An Interference-Minimized MultiPath Routing protocol (IM2PR) example is given in [64], which aims to find a sufficient number of minimum interfering paths with high data transmission quality in event-driven WSNs.

Frequency-hopping As the ISM radio band contains limited channels, many wireless networks protocol have to co-exist within this band, including ZigBee, Wi-Fi, UWB and Bluetooth. Additionally, if a network has only one working channel, it usually suffers from frequency interference and packet errors. However, with the increasing development of wireless communication, wireless networks can operate at higher frequencies, especially with the emergence of 3G, 4G, and 5 GHz technologies and related architectures. WSNs can resist interference between different protocols and avoid network congestion by operating within multiple channels and using frequency-hopping. Experts have proposed many novel strategies to increase the reliability of WSNs. In [105], the authors proposed MMSN, which takes advantage of the multiple frequencies available, and considers the restrictions of WSNs. In MMSN, four frequency assignment options are provided to meet different application requirements by frequency-hopping. In [73] an adaptive frequency hopping method has been proposed, which triggers frequency hopping only when the packet drop rate of the channel exceeds a given threshold.

3.1.3 Longevity

In a wireless network, where most wireless sensor nodes are battery-powered, the replacement and recharging of a large number of batteries is impractical and carries a higher cost. Therefore, the rest energy of nodes is relevant to the lifetime of the overall network. As increasing numbers of WNs are being applied and mobile devices are being added to networks, nodes not only collect, fuse, and transmit data, but also move independently and perform other mechanical operations. Consequently, in this part we will focus on these two aspects to prolong the operation of IWSNs and discuss the longevity.

Power-efficiency In recent years, many research efforts have focused on power-efficiency within application specifications [66]. Actually, each activity performed by nodes in a WSN consumes energy, and power limitations, efficient energy management and conservation are vital measures to ensure longevity. Energy efficient strategies have been implemented within different layers and functions. Firstly, within the physical layer, unnecessary actions can be reduced and the physical parameters can be optimized to achieve strong power-saving performance, especially for mobile nodes. For example, in order to meet coverage and network connectivity requirements of WSNs, the authors in [50] have formulated a mobile sensor deployment (MSD) problem with the aim of deploying mobile sensors to provide the target coverage and network, and then increase the battery lifetime by minimizing the nodes’ movement. Secondly, within networking and communication, the protocols of MAC, routing, and transportation have been modified to reduce the energy consumption. For instance, in [24], a sleep and working mechanism is introduced for communication to achieve good energy conservation effects. Finally, within the application layer, there are several useful methods that can be used to decrease energy consumption such as event-driven techniques, application-driven techniques and efficient data/messaging.

Energy-harvesting Energy-harvesting can be realized by transforming ambient energy into electric energy and transferring power wirelessly [42]. Energy-harvesting systems include harvesting devices, power management, and energy conservation strategies. Many researchers have study this issue by exploring other environmental energies such as solar, wind, water flow, vibration and thermal energy. It is obvious that these studies can be used as references when WSNs are applied in industrial domain. Table 5 summarizes possible sources of energy for harvesting in more detail.

Table 5 Energy-harvesting for WSNs
Full size table

3.1.4 Privacy and security

For wired networks, an intruder requires a physical connection when trying to hack into the network. However, wireless networks have a higher potential for intrusion, especially for wireless networks operating within the ISM band. Furthermore, more public wireless/wired networks are beginning to be merged with IWNs, for example industrial Wi-Fi and civil Wi-Fi have similar protocols. This ongoing integration increases flexibility, convenience, and scalability; however it introduces adverse risks for security and privacy [18, 60]. In general, IWSNs are more susceptible to security attacks and privacy leaks. Consequently, new security and privacy-preservation technologies have been developed for wireless networks or IWNs. The following paragraphs will briefly discuss privacy preservation and the security of WSNs from an industrial perspective.

Privacy preservation Within the Industry 4.0 framework, enterprises can provide customized services and goods publically using data mining and artificial intelligence perdition techniques. The private data and information of individuals will be collected and uploaded to the cloud by hybrid networks. This information may be exchanged between different intelligent devices, workers and so on. Privacy preservation is therefore becoming a greater challenge within this situation. In [59], the author has analyzed existing approaches for privacy protection in WSNs from two layers, data and sensor. Furthermore, the privacy preservation approaches applied in large-scale industrial environments has been discussed. Privacy preservation can usually be divided into two categories, data and context protection. Data protection includes data aggregation and data querying, while context protection focuses on protecting the contextual information, such as the location, timing and information transmission in a WSN [54]. For privacy protection, access control and authentication plays an important role in protecting security-sensitive sensor data from being utilized by malicious users [48].

Security It is apparent that wireless networking security has become an important topic in recent years [30]. Furthermore, within the IWN 4.0 framework, the internet or other communication network is used to connect all devices or components. Open access IWNs will face both traditional security threats, as well as other attacks from the internet, especially for wireless networks. Defense mechanisms are briefly surveyed and listed in Table 6. A complete list and more information can be found in [30].

Table 6 Attacks and coding methods
Full size table

However, due to the differences between wireless networks and wired networks, and the special requirements of industrial application, traditional security measures may not be suitable for IWNs. To deal with these emerging problems, some novel strategies have been developed [3]. There is valuable experience to be taken from these IWN studies; however the industrial environment and the specific IWN requirements are not considered by these studies and therefore there is a requirement for new security and privacy methods to be proposed in future.

3.2 QoD in IWNs

Data quality is an important factor for wired networks, especially for the internet. However, since most of the data in traditional IWNs is collected to solve special problems, monitor parameters, and deliver machine instructions, the overall system does not have a requirement for high quality data or information. However, for Industry 4.0, almost every workflow relies on high quality data and precise information and data is becoming a primary component of a successfully operating industrial system [14]. In actual smart factories and systems using Industry 4.0, every device, worker and piece of equipment can be considered to be a data client, and all operations are supported by the industrial cloud and data mining engine which provides higher productivity [102]. From the perspective of the user, IWN data should have guaranteed validity, accuracy, reliability and integrity, as shown in Fig. 5. Therefore, we will analyze these data characteristics and attempt to provide some data quality indexes.

Fig. 5
figure 5

QoD for IWNs

Full size image

3.2.1 Validity

As cloud and communication technologies (such as 3G and internet) are becoming more widely available, more and more data is being produced. The data quantity unit is rising from MB, GB, and TB to PB. Experts, users and researchers are facing the acute problem of how to retrieve useful information from the massive ocean of data. In Industry 4.0, factories, users, devices and IWNs are faced with the same big data issues. This massive quantity of data will have several effects. Firstly, growing data volumes will make it more difficult to mine valid information. Secondly, massive data will result in heavy traffic loads for IWNs. Last but not least, it requires more resources to be devoted for storage, analysis and maintenance of this massive data.

To address this problem, data validity should be improved for a more efficient solution. IWNs can realize this by introduction of cross-layer enhancements. Firstly, validity of the data collection requires improvement of the physical layer, from sensing to gathering. Secondly, for valid data acquisition, data fusion should be considered, such as setting valid data constraints. Finally, the validity of the data directly impacts the performance of the IWNs. Thus, more valid information can be obtained from the application layer such as within industrial clouds, where data is stored and computed using optimized algorithms.

3.2.2 Accuracy

Accuracy is an important data quality index. In the industrial domain, data accuracy is directly related to the accuracy of the machinery and equipment workflow, such as the assembly, production, logistics and other links. Additionally, data accuracy can have an impact on the quality of the products, the production efficiency and the economic profits. In an IWN, the context can include vital information about the data itself such as the time, hop number, neighbors, and hidden bugs of the system can be found by data mining.

3.2.3 Reliability

Reliability is a different but no less important data index, and it assesses the stability and consistency of results of the system or equipment. The reliability of data can reflect the performance of the whole system, especially IWNs and industrial cloud services. In IWNs, the delivery of reliable data is susceptible to the mobile and dynamic topology. An unstable data stream can result in a link failure or packet error. Therefore, reliable routing and transport protocols will contribute to the reliability of data that echoes the QoS of IWNs. Effective strategies should consider the effects of data collection, routing layer and congestion control. This will be discussed in further detail in Sect. 4.1.

As discussed above, data reliability affects the whole system architecture, from the physical layer to the application layer, and it can be concluded that reliable data not only depends on reliable transportation and collection, but also depends on computing and analyzing results from clouds. Many IWN users, especially workmen, dealing with some type of alarms, and automated production steps rely on analysis of results from the raw data. It can be imagined that failure of the analysis system to provide reliable data/results can cause unexpected situations to occur for the whole factory.

3.2.4 Integrity

For Industry 4.0 and IWN applications, it is easy to obtain incomplete or deficient data due to packet error or loss, transportation failure, and data limitations, particular in multiple object systems and large scale networks. It is well known that incomplete data can be harmful to Industry 4.0 and the smart factory. However, many studies of IWNs, industrial clouds and applications are undertaken based on the assumption complete data which may not be reflective of real situations, where engineers have to cope with incomplete and deficient data. The issue of how to guarantee data integrity is becoming a key research area for Industry 4.0 and IWNs.

The four indexes described above of integrity, reliability, accuracy and validity are closely related to each other. However, validity can be considered to the most important index and it is dependent on the other three measures.

4 A taxonomy of IWNs applications

Industry 4.0 is a revolution for the whole of industry and society, and IWNs are an important component within its framework. Current studies, technologies, concepts and frameworks are the basis of successful implementation, realization, and application of Industry 4.0. Existing standards, applications, and products of IWNs need to be reviewed, since the success of IWNs directly contributes to Industry 4.0. In this section, we provide a brief comparative study of emerging and existing IWN standards such as ZigBee, Wi-Fi and other maturing technologies. The taxonomy of IWN applications from monitoring and automations are then discussed. We also survey some popular products and available IWN devices.

4.1 Current standards and products

Standards Due to the importance of IWNs, they have been the focus of many research groups. These basic conditions and special requirements of study and market have propelled designers and alliances to propose various standards for IWNs. At present, there are several main streams of standards for IWNs based on IEEE 802.15.4 and 802.11 [14, 18, 51, 60].

In Table 7, we list five IWNs standards: WirelessHART, ISA100.11a, WIP-PA, ZigBee, and Wi-Fi. A comparison of some parameters such as the physical layer, MAC protocol, working frequency, and maximum data rate are also listed. It is evident that most standards share the same framework (IEEE 802.11.15.4), and radio frequency. As IEEE 802.11 research progresses, Wi-Fi is becoming another main option, as a result of the ease of access to the internet (and other networks), and connecting to mature and extensive applications. However, Wi-Fi is not designed for industrial applications; so many adaptations to Wi-Fi are required to meet industrial requirements, at both a protocols and device level.

Table 7 Comparison of different standards
Full size table

IEEE 802.15.4 based protocols can be utilized in scenarios where the unit node cost and power consumption are limited. Due to the low data rate (250 Kbps), these protocols can only provide limited QoS. In contrast, Wi-Fi can provide a more suitable QoS for most applications with an air data rate above 10 Mbps with a higher investment. Additionally, IEEE 802.11a/b/g only supports a star networking topology while IEEE 802.11n/ac has started to support wireless bridging services to hold tree topologies. Industry planners should be aware that the mesh networking topologies which are common in IEEE 802.15.4 are not well supported by IEEE 802.11 standards.

Products At present, many enterprises such as Siemens, MOXA, Cisco and Advantech have developed a series of industrial wireless products, services and solutions, based on the advantages of wireless networks. These wireless networking products can be divided into three classes: access points (AP), client modules and bridges (or relays), especially for industrial Wi-Fi. Most of these wireless products are based on IEEE 802.11x and IEEE 802.15.4 frameworks and private protocols. Additionally, the main enterprises have given some improvements for adapting industrial applications based on the different applications.

Table 8 compares the features of various representative products in terms of company and data rate. Although there are additional indexes of IWN products available, we also focus on other important factors, such as the company, wireless standard used, maximum data rate, and function within the network. These products’ features reveal the main characteristics of a product from the general application designer’s aspect. It can be seen that almost all products achieve a high data rate, however some possess low data rates with the full range from 250 Kbps to 1.3 Gbps. It is obvious that this is sufficient for large scale IWNs or applications.

Table 8 A comparison of IWN Products
Full size table

As discussed in previous sections and shown in Table 8, Wi-Fi (IEEE 802.11) can provide a higher data capacity with a higher investment and restricted networking topologies (tree and star). IEEE 802.15.4 solutions are suitable for scenarios where the unit cost and power consumption are limited and data capacity is not the main concern. Meanwhile, Wi-Fi can easily access the internet for low cost and devices. Private protocols are not encouraged within the open interconnection requirements of Industry 4.0.

4.2 Comparative studies of existing applications

It is obvious that wireless networks have more advantages in most industrial applications than wired networks. Furthermore, under the concept of Industry 4.0, growing numbers of mobile devices need to communication with each other, such as M2M. Additionally, moving objects raise challenges and constraints for wired networks. These factors have all led to wireless networks becoming the first choice for most applications. As communication develops, wireless networks are being applied in a wide range of industrial fields and other domains such as health care and agriculture. In this section, we will focus on discussion of the industrial applications of IWNs. Although there are many IWN application scenarios such as refineries, cement plants, chemical plants and automobile factories, all IWN applications can be divided into two categories: parameter monitoring and automation control. The following is a compilation of the most relevant projects proposed in recent years in the field of IWNs, as illustrated in Table 9. We focus on the comparison indexes of IWNs for topology, protocols, medium and target application. Note that N/A means “not applicable” in Table 9.

Table 9 A comparison of existing IWNs projects
Full size table

4.2.1 Parameter monitoring

Monitoring is an important traditional application field in the industrial domain. Wireless sensor nodes of IWNs gather different information from sensors, and then upload safety or alert information to staff or management applications wirelessly. Generally, there are two important applications for monitoring systems.

Environmental monitors the industrial environmental condition features (such as temperature, humidity, dust index, fire alarm, toxic gas and pollution) of the whole factory and other plants.

Study [101] has provided a good case study for the energy industry and has presented an integrated environment monitoring system for an underground coal mine. The overall system includes a cable monitoring system (CMS) and multi-sensor IWNs and the wireless nodes were adapted to monitor coal mine parameters such as gas, smoke, and the temperature of the coal mine environment, in order to avoid accidents. For successful implementation of the project, two work modes were adopted: periodic inspection and interrupt service. Other mechanisms for dealing with topology, location, energy, and fault management have been proposed.

In [33], a sensor system network for remote monitoring of the state of pollution using high-voltage insulators has been designed, deployed, and installed on 230- and 500-kV transmission power towers in Northeast Brazil. In this system, data is sent wirelessly to a satellite, and then transmitted from the satellite to the database servers.

Conditions The second main monitoring application monitors equipment operating conditions and system state such as noise, vibration and other features. Workers or the system may detect equipment trouble or failure, which can be used to prepare for repair, maintenance, and replacement.

The SWATS [95] project is an example of this type of application for monitoring equipment conditions. The authors have designed a system with wireless networks and sensors to obtain information by monitoring an oil pipe in an oilfield. In the system IWNs detect, identify, and localize major exceptions such as plug leakage, generator malfunction, blockages and other problems within the oil field pipe.

In [38], another good example is given which illustrates an industrial wireless communication system for monitoring equipment status. In this paper, the motor stator current and vibration signals are measured by IWNs containing special sensors. The IWNs transmit the sensing data to servers. After processing and analysis, staff managing the data can obtain motor condition information. The final results can be used within an expert system to complete the fault diagnosis for the machine.

4.2.2 Automation control

One of the main applications of IWNs is automation control systems, especially for transmission of instructions and essential information from the server or industrial cloud. In Industry 4.0 systems, there can be interactions of information within devices, between devices and the server, worker and clouds.

Systems without sensors An automation control system can be constructed using wireless nodes, RFID, and automation devices such as a programmable logic controller (PLC) and other controllers. In these applications, industrial wireless nodes without sensors are responsible for transmitting and receiving periodic instructions. Automated devices perform their own tasks such as moving, and rotating, especially in process control systems. In [88], a master–slave wireless industrial system was designed to realize production automation in a cement mill. A wireless module (SY-S72) operating at 433 MHz was mounted on an industrial computer and PLCs (S7-200). The slave nodes directly control the automation equipment such as the miller, transporter, and cement kiln. The master transmits the commands to the slave nodes using wireless modules. These wireless nodes have high-gain antennas for remote wireless communications. The PLC of the slave nodes finish the automation control based on the received commands.

Since IWNs have more advantages such as reducing the labor force and increasing flexibility and mobility, wireless communication technologies can be used to organize thousands of goods in a warehouse, and is a good project for this type of IWN application. In [89], AGVs, RFID, wireless networks are adopted to finish categorizing each product, in order to save working time and resources and avoid wired network constraints. Within this project, wireless networks communicate among AGVs, between AGVs and products, and this project has shown stable, reliable and seamless application of IWNs.

Systems with sensors In these systems, different sensors are connected to wireless nodes to obtain vital data and transmit this data to the actuator. The actuator uses the sensing data to complete the next step of the automation control. In this type of application, wireless nodes have two roles: sensing data and transmitting data. The actuator and wireless sensor nodes form a closed loop. WSNs and WSANs in automated system applications are good examples. In [6], a control system was proposed that contains a programmable logic controller (PLC), an industrial wireless local area network, and sensors. A pressure transmitter (sensor) was used to sense the tank water level. When the water level exceeds the limit, the wireless node transmits a signal to the PLC to control the water pump.

5 IWNs design: a case study

In this section, we will use one of our developed IWN projects in the context of Industry 4.0, clouds and big data to explain the implementation steps, the main configuration and functions of IWNs as well as the whole system working principle. The case is an IWN design prototype for a smart factory and Industry 4.0. More information can be found in [80].

It is known that the goal of the Industry 4.0 and smart factory frameworks is to establish a new platform and bridge that will support the manufacturers and users, and realize wireless communication among smart devices, workers and users while offering the advantages of flexibility, no wiring, and mobility. Thus, our smart factory prototype is based on IWNs, the cloud, smart machines and a management system. As shown in Fig. 6, this prototype contains five main parts: (1) raw products with RFID tags, industrial environment smart sensors, smart machines and robots, (2) a conveyance system, (3) an IWN, (4) a private industrial cloud and (5) smart terminals.

Fig. 6
figure 6

A case study of IWNs under Industry 4.0

Full size image

The working principle of the overall prototype consists of several steps. Firstly, users customize their products of interest using the web page, and then related information about the user and the products are uploaded to a server. Once the raw information is obtained, the private cloud then builds a machine processing program, process, and resource utilization strategies, which are issued to the smart machine, workers, and conveyance system to finish the product manufacturing by IWNs. At the same time, manufacturing data and information about the user’s interests are transmitted to the industrial private clouds. Additionally, the related data and graphics are distributed to terminals for management and user validation and decision-making.

In this case, there are two flow systems supporting the prototype system under normal operation. Firstly, from a physical view, the conveyance system is responsible for transporting the products. Secondly, from an information view, IWNs play an important role in the whole system. In contrast with traditional industry systems and factory, huge amounts of data will be produced during processing, and more and more information needs to be transmitted and processed by IWNs under the new challenges of Industry 4.0. Information is the basis to perform tasks for the whole system. However, multiple devices, network interface and communication protocols between the raw product, machines, workers and robots are needed to converse and communicate. As discussed above, IWNs provide the bridge and information communication. They are the foundation and key for realizing Industry 4.0 and smart factories.

6 Issues and challenges

Communication network design goals differ, due to the variety of real industrial domains [35]. In this section, we will discuss the recent research trends of IWNs from a QoS-oriented perspective, based on our IWNs design experiences as in Fig. 7. This figure summarizes the challenges and corresponding QoS indexes for IWNs. There are still some significant challenges that need to be solved to meet the diverse IWNs application and QoS demands before IWNs technologies can be widely applied.

Fig. 7
figure 7

QoS and research issues

Full size image

Reliability and topology control In IMWSNs, the network consists of hybrid (mobile and static) nodes or only mobile nodes instead of static nodes. Therefore, it is necessary to present an effective way to deal with problems such as dynamic topology and reliability caused by the mobility. Note that topology control is one of the key design goals for routing protocol design to meet the reliability of the application [53, 103].

Error tolerance It is known that the error tolerance directly impacts the success of communications. This is particularly relevant in an industrial environment where the challenge of signal interference and path loss are more significant which will increase communication errors. Thus the overall IMWSN should establish an error tolerance mechanism. Meanwhile, error tolerance is important to guarantee the adaptiveness and robustness of the IWN [8, 81].

Quick handoff When a WSN contains mobile nodes in an industrial scenario, the designer needs to consider quick handoff for real time and seamless handover, which may be required between different multiple access points, neighboring nodes, the base station, or sink nodes [98].

Node deployment/placement Generally, node deployment/placement schedules are designed for static nodes in conventional WSNs, and mobile nodes were not considered [78]. Taking this into account, the deployment strategies for WSNs are therefore unsuitable for IWNs and the industrial application and environment must also be considered. Therefore, some operational node deployment algorithms have been adopted to solve this new challenge.

Communication protocols By adding mobile nodes, the whole mobile network becomes more flexible and complicated. From the perspective of communication protocols and industry-specific applications, some new feature requirements arise. As a result, new MAC, routing, transport and application protocols should be proposed to meet these requirements [18, 35].

Efficient energy management For Industry 4.0, energy consumption minimization and efficient energy management is the key to producing products and prolonging the network’s lifetime [43, 70]. As shown above, in contrast with WSNs, mobile nodes are powered by batteries. Therefore an efficient moving path and communications become vital design goals.

Data fusion In Industry 4.0 systems, users, factories, and equipment are linked together. For data-based industrial wireless sensor nodes, data validity becomes an important factor. In the Industry 4.0 framework, all components including mobile nodes, are required to be intelligent or smart devices, so that all mobile nodes can process raw data and fuse this into useful information. Data fusion can meet different industrial requirements as it is evident that there is sparse information contained within large quantities of raw data. The ratio of useful data will be increased by data fusion, while latency and energy consumption will decrease by reducing the amount of data [97].

Self-adaptive operation One of the greatest differences of WSNs is that in an IMWSN, the mobile node has self-adaptive operation for its location, motion, communication, and other functions. Therefore the mobile node is intelligent and can adapt to complete its tasks [79].

7 Conclusions

IWNs are an important and effective way to improve flexibility, productivity and networking of an industrial system by providing greater mobility, intelligence, and adaptation. IWNs are a promising technology which will play an increasingly important role in the next generation of industrial systems for Industry 4.0. However, there are few IWNs surveys that consider the background of new technologies, such as industrial clouds, big data, and the context of Industry 4.0, so this is motivation of this paper. IWNs create a new set of challenges in terms of real-time, reliability, longevity, security and privacy, data quality and other QoS indexes. We have presented a brief survey of Industry 4.0, IWNs and wireless nodes covering a range of areas from the general to the specific. A QoS and QoD-oriented architecture, existing solutions for QoS, and QoD are discussed in detail. We have also summarized the current main standards, applications and products for IWNs. Despite advances in these areas, there are many challenges that still need to be addressed within the Industry 4.0 concept, especially for topology control, signal interference, communication protocols, hybrids between IWNs and other wireless/wired technologies, and the design of successful applications.