of IIoT systems is a major issue for smart environments and services. In designing a system, we need to actively evaluate system performance and its efficiency. Multiple factors such as outage probability, power consumption, and spectral efficiency could verify the performance and efficiency of IIoT wireless design. These metrics can be categorized into two broad classes, namely, resources and services indicators.
1.6.2.1 Resource Indicators
Wireless solutions in IIoT should specify the required resources to create connectivity within the systems. To identify the level of network resource utilization in wireless systems, a set of metrics are introduced as resource indicators: spectrum, time, power, and network computing [55].
1 Spectrum: In the frequency domain, the wireless channels are identified by the channel bandwidth. Given that the data rate of each wireless channel is related to its bandwidth [184], different services in IIoT environment are highly dependent on spectrum resources to accommodate the desired data rate. For instance, in mission-critical communications such as safety systems, the radio spectrum resource should be assigned properly for effective deployment (i.e., high data rate and low latency) [185]. Given that spectrum resources are shared in wireless domain for a large variety of applications, wireless networks suffer from spectrum scarcity. This highlights the need for communication protocol interoperation within different frequency bands and channels [186].
2 Time: Wireless spectrum resource usage could be defined in terms of the transmission time and length. Time manages the temporal operation of wireless resources based on the desired performance such as QoS [187]. In this context, the access time of channel spectrum could be considered a resource indicator to be optimized for wireless networks.
3 Energy consumption: Power consumption of wireless nodes at transmission side is closely dependent on transmission range, quality of transmission channel, and transmission power at RF front ends. The aforementioned resources are essential in the design of a wireless system. Various energy-efficient resource allocation techniques (with given QoS constraints) are proposed to assist in design of wireless systems for IIoT.
4 Computing resources: Network resources are in both hardware and software types (physical and virtual). Basically, computing resources include all network resources tied up to the aforementioned resources (i.e., spectrum, time, and energy consumption). For instance, in TDMA-based transmission, different communication links are assigned to users at different time slots and through computing resources.
To achieve the diversity gain for different applications in IIoT, the resource space is used in multiple dimensions. For example, a wireless solution could exploit a TDMA scheme in the multi-antenna wireless systems, while both power and spectrum are considered in the system design.
1.6.2.2 Service Indicators
In addition to wireless network resources that specify the cost to build connections, service performance highly affects the design of IIoT networks. Several service measures are identified to indicate the level of service satisfaction (while ensuring QoS), and they are extensively utilized in formulating wireless design. Service performance factors that form a set of basic metrics are data throughput, delay, coverage, and scalability.
1 Data throughput: Service requirements are specified by several data rate metrics such as real-time throughput, aggregated throughput, raw data rate, good throughput (goodput), peak and minimum throughput, average throughput, and instantaneous data rate. Throughput is the key indicator for high-performance services such as autonomous driving and robot-assisted surgery.
2 Delay: It specifies the latency over the communication link and is composed of the transmission delay, queuing delay, and jitter. As a service indicator, delay defines the speed of data delivery across networks and verifies the timeline for data packets transmission. This is an essentially important metric for mission-critical industrial applications such as real-time control and safety that are featured by stringent latency requirements.
3 Coverage: It is defined as the wireless network range that connectivity is provided for services. Generally, the worst-case channel for all terminals in the coverage area is considered as the performance metric of wireless network.
4 Scalability: It stands for the number of nodes that could be connected to the wireless network. Given that there are various IIoT services with different levels of scalability, this performance metric is highly associated with the traffic load of network.
Generally, the combination of multiple indicators forms a composite service metric in wireless systems, and the selected indicators are optimized for the given service requirements.
1.7 MAC Protocols in IIoT
Industrial environments have some special properties such as equipment noise, electromagnetic interference, and complex coordination among devices that pose unique challenges for wireless connectivity. IIoT systems often require support of seamless communication and diverse objectives of functions for a wide range of applications. Therefore, wireless communication protocol design that meets these performance criteria is essential to fulfill service QoS and ensure prolonged lifetime.
MAC layer is a key sublayer of the data link layer that exploits use of specific protocols to administer nodes privilege to access shared wireless medium according to application requirements. MAC protocol controls the radio and channel sensing scheme and defines nodes duty cycle, communication mode between devices, data rates, transmission power, and range. Given that radio is the foremost source of power consumption in networks, MAC protocols could significantly impact nodes’ overall power consumption and their lifetime [188]. In addition to the aforementioned key features, the MAC layer also controls other wireless settings such as frames synchronization, source–destination address management, error detection for physical layer transmission, collision reduction, and mitigation of idle listening.
With the goal of meeting the requirements of IIoT applications, the existing protocols could be adapted, evolved, or developed for different performance characteristics. MAC protocol schemes can be categorized into two broad classes: scheduled based and contention based [189]. Hybrid schemes are also proposed as a combination of these schemes.
1.7.1 Scheduled-Based Schemes
In scheduled-based protocols, also known as fixed reservation–based schemes, a fixed duration of time, frequency, or other domain is scheduled and assigned to nodes for network resources access. The scheduling assignment algorithm is conducted by a centralized base station and aims at avoiding channel collisions. In addition to a collision-free schedule, a device is simply set to sleep when it is not using its time slot to prevent idle listening and message overhearing [190]. This scheme is more suitable for networks that deploy low-mobility nodes and require infrequent topology changes and scheduling adjustment. It also tends to be more predictable and offers deterministic E2E delay. However, in dense networks, nodes should wait to gain access to the wireless medium, and additional queuing delay shall be incurred. Synchronization is an important issue in this approach leading to higher complexity and additional traffic due to additional control packets. The following multiple access schemes are utilized in typical multi-user wireless communication systems.
TDMA: Time is divided among nodes for a given and identical frequency channel. Therefore, a fixed portion of time is assigned to every node to transmit data. For successful TDMA slot assignment and collision-free communication, tight clock synchronization should be established between nodes. GinMAC [191] and wireless arbitration (WirArb) [192] are some collision-free TDMA-based MAC protocols in time-sensitive IIoT. For instance, GinMAC provides reliable data delivery as well as deterministic time delay for industrial process automation such as closed-loop control systems.17 Given that only one node is allowed to transmit data during the scheduled time slots, TDMA suffers from relatively high delay.
FDMA: As the name infers, accessible frequency bandwidth is partitioned into non-overlapped sub-channels, where each individual sub-channel is adequate to accommodate transmission of a signal spectrum. Ideally, through proper