Reconfigurable intelligent surface with 6G for industrial revolution: Potential applications and research challenges

3.1 Coverage enhancement

Due to significantly higher bandwidth at the mmWave/THz wave regime of the spectrum, it has potential to support high data rates. However, wireless communication at these frequencies suffers from severe path loss. This potential communication problem could be mitigated by using a massive array antenna gain/directivity within a compact space. In addition to this, another impediment, that is, vulnerability to the blockages by obstacles, could be rectified by the deployment of RIS to accomplish an ancillary communication link even when the direct link is obstructed. The RIS could be installed to develop an artificial link between the base station and users who experience weak received signal strength or obstruction [9]. This application also proposes a potential resolution for the enhancement of the coverage, particularly for the communication with the mmWave/THz wave, which is highly susceptible to the obstruction.

3.2 Concurrent information and wireless power transfer

The RIS is employed to exploit its potential features including beam-forming/beam-focusing as well as beam-steering proficiencies toward the devices consume excess power to fulfill the objective of both the transmission of information signal and power simultaneously [10]. This feature of RIS could be a promising outcome for evolving research directions. However, the crucial task in “simultaneous wireless information and power transfer” techniques is that of the energy receivers (ERs) and information receivers (IRs) which operate under different power supply constraints, with the ER receiving power at a significantly much higher order than IR. Therefore, the ER must be implemented nearer to the base station than the IR to yield adequate power because the signal reduction restricts the ER’s applied functioning range.

3.3 Physical layer security

The physical layer of wireless communication links is highly susceptible to protection threats. Thus, the security concerns at this layer have received wide research consideration because they must prevent complicated key communication protocol and be appropriate for latency-susceptible applications. The RIS is installed to deliberately depreciate the signal perceived by eavesdroppers which can be obtained by adjusting the RIS unit cell’s phase shift as a reflected link through which it is superimposed destructively to the signal of the legitimate user at the eavesdropper’s receiver, which must reduce information outflow [11]. Further, to extend the rate of a secure communication link, both artificial noise and multiple antennas have to be incorporated. However, the achievable secure rate remains limited when both the appropriate users and eavesdroppers have correlated channels or when the eavesdroppers are nearer to the base station than the authentic users. In addition, cybersecurity perspectives that are yet to be explored are going to be critical in the near future [12].

3.4 Positioning

Currently, radio/millimeter wave localization has been increasingly considered as an emerging technology. With this context, the RIS plays a very significant role and enables high-precision positioning of users, dependent on the “time-of-arrival and departure” measurements. With reference to the nature of users, the localization methods could be classified into two major classes: 1) active method and 2) passive method. In the active case localization, the user transmits and receives the signals; however, in the passive case, the user only reflects/scatters the signals from a transmitter. The narrow beams that could be created in 3D-space using large RIS are exploited to estimate the position of terminals precisely in this use case as reported by Hu et al. [13]. Due to the high sensitivity of satellite signals to obstruction, the traditional positioning (i.e., Global Positioning System) typically fails; therefore, RIS-supported radar could be a promising use mainly for indoor localization. Further, a positioning algorithm with low complexity that applies orthogonal sequences in the design of the phase profile of RIS needs to develop which can resolve the multipath interference and the problems associated with data of the received signal. Ultimately, the localization error of the used approach needs to evaluate which illustrate theoretical “Cramer–Rao lower bounds”.

3.5 Rank enhancement

Distributed RIS deployment should work as engineered scatterers and integrate a sort of multipath propagation such that additional degrees of freedom are established which is able to solve the rank deficiency problem [14]. For example, the channel matrix of MIMO turns out to be rank deficient in the LoS environments; therefore, the spatial multiplexing becomes unattainable.

3.6 Support cell edge users

RIS deployment on the cell edge users could be served for the users who generally endure from both high signal reduction from the base stations and high intervention from the neighboring base stations. To develop an interference-free region, at the intended users, the reflected signal from RIS is superimposed constructively to create a signal hotspot and destructively at the inadvertent users [15].

3.7 Support massive D2D communications

To establish an efficient and intelligent communication among numerous low-power devices with the deployment of RIS in the device-to-device communication environment play a significant role and work as a reflection hub where the reflected signals are superimposed constructively at the authentic users and destructively at undesired users [16].

3.8 Support modulation

An interesting application of the RIS is to provide support as an access point to establish an efficient communication. The potential features of RIS have been exploited as a dedicated RF source for information encoding due to its potential features in the reflected signal such as its amplitude, phase, or polarization [17].

3.9 Support multicell networks

Multiple base stations in various communication cells typically share limited frequency resources in order to maximize “spectral-efficiency,” which causes “inter-cell interference,” especially for “cell-edge-users.” The fundamental signal power obtained by the cell edge user from its serving base station is comparable to the interference anticipated from its nearby cells. Consequently, the users at the cell edge experience poor “signal-to-interference-plus-noise ratio”. To solve this problem, Pan et al. [18] suggested deploying RIS near the cell boundary, which will be able to simultaneously improve the signal received from the active base station and reduce interference from other users. Additionally, according to the simulation results shown by Pan et al. [18], the sum rate achieved by RIS-supported systems is about twice as high as that of systems without RIS.

3.10 Support mobile edge computing (MEC) networks

For the emerging computationally intensive applications such as video-processing tasks, computation-intensive image processing, and immersive technologies would be executed in real-time scenarios but these tasks could not be accomplished locally due to the limited power supply and hardware capabilities. Thus, these computationally intensive errands can be distributed over the robust computing nodes which are generally deployed at the edge of the network. Nevertheless, as a special case when these devices are far from the MEC node, they suffer from a low data-distributing rate due to terrible path loss and results in offloading delay. With reference to these issues, a novel RIS-supporting MEC framework is proposed by Bai et al. [19] to surmount this obstacle. The phase shift, discrete amplitude, mutual coupling, and hardware impairment need to be considered in the practical implementation of RIS hardware architecture. 6G communication architecture with RIS is a very appropriate approach for high-speed vehicular communication systems because it provides virtual LoS path loss with less energy consumption to enhance the effective channel power gain. In addition to this, it also provides a better solution to solve the problem of Doppler’s effects caused by the mobility of high-speed vehicles.

3.11 Support nonorthogonal multiple access (NOMA)

NOMA is a multiple access future generation communication technique in which each orthogonal resource block is shared by multiple users simultaneously, which enhances the spectral efficiency. However, NOMA is not a better option when the users’ channel vectors are orthogonal to each other. Thus, to expand this application, RIS is introduced into the system to manipulate the wireless channel vectors for all the users so that the user’s channel vector could be aligned with another one’s [20].

3.12 Support multicast networks

Multicast transmission has attracted wide research attention relying on content reuse because it mitigates the tele-traffic and plays a crucial role in wireless networks. In “multigroup” and “multicast” communications, identical content is shared within each group, and each group’s data rate is limited to the user with the weakest channel gain. The RIS-supported multicast architecture is described by Zhou et al. [21], and this problem is tackled by meticulously tuning the RIS phase shifts to enhance the channel conditions of the weakest link.

3.13 Support CR communication networks

CR is a communication device which is capable of enhancing the spectral efficiency by permitting to reuse the underutilized/unutilized spectrum of the primary user by the cognitive user without interfering or with tolerable interference to the primary user. Thus, in “cognitive radio communication networks,” the interference management between users, particularly primary user and cognitive user, is a very challenging task to enhance the system performance. With this context, the RIS plays a very important role in mitigating this issue. RIS provides a method to engineer the wireless propagation ecosystem and it can engineer only the reflection paths. However, the RIS is unable to completely cancel the existing interference if the direct paths between the transceivers are very robust, but it helps to reduce the interference.

3.14 Support industrial revolution

Industrial revolution, that is Industry 4.0/5.0, has been powered by robotics and wireless communication technologies and performs a very decisive role in the digital evolution of industrial developments, warehousing, and logistics. Real-time manufacturing applications, such as immersive technologies (“augmented and virtual reality”) to assist industrial performance, may require giga-bits-per second (Gbps) peak data rates [22]. As a potential solution, the mmWave/terahertz wave has been identified as a supporting wireless communication network in the industrial scenario [23,24]. Nevertheless, the communication over high-frequency band suffers from extreme obstruction sensitivity as reported by Jain et al. [25], which reduces communication consistency when an automated machine moves in the industrial environments with obstacles, and the trajectory scheduling will highly affect the mmWave communication performance as it controls whether the machine is in LoS or “non-LoS” to avoid obstacles. Furthermore, robots are battery oriented and typically assigned jobs with strict timeframes. By maximizing the robot’s movement, its energy consumption could be significantly reduced along with a large drop in the overall electrical energy used for production processes. As a result, one of the most important issues in robotics is the robot trajectory planning [26]. For wirelessly connected robots, where the trajectories must be tailored to the radio coverage, this issue has grown in significance [27]. The minimization of motion energy for a robot that uses mmWave communications and is supported by an RIS is a critical issue for fully automated industries controlled by the combination of autonomous robots and new generations of mobile communications (5G/6G) [28]. Additionally, the communication dependability and the energy efficiency of the robot continue to be crucial issues that must be addressed in order to optimize robot trajectory and communication QoS. As a result, it is necessary to optimize beamforming at the RIS with access point, robot trajectory, communication QoS, and robot position in order to ensure mutual reliance. According to Tatino et al. [28], passive RISs are an effective way to increase the radio coverage and “motion-energy-efficiency” of robots.

4.1 Channel estimation

To obtain the full advantages of RIS-supported wireless networks, the CSI has to be estimated correctly to support the phase shift design. For this purpose, we can consider instantaneous CSI (i.e., short-term CSI) which is the current channel condition that offers an opportunity to adapt the transmitted signal and thus optimizes the received signal for spatial multiplexing to accomplish low bit error rate. Additionally, statistical CSI (long-term CSI) data offer statistical channel characterization that could be applied for transmission optimization. Furthermore, because of the channel situation, CSI procurement is virtually constrained. Only the statistical CSI is appropriate in a fast-fading environment since the channel conditions change quickly during the transmission of a single information symbol. The immediate CSI, however, may be computed with tolerable precision and used for transmission adaptation for a while before becoming out of date in the slow-fading conditions. In the real-world circumstances, the accessible CSI frequently falls between the statistical and the instantaneous CSI, and the information from the later CSI is integrated with the information from the earlier CSI that has estimation/quantization error. We have considered an RIS-supported wireless network where a multi-antenna base station serves a single-antenna user with RIS for near-instantaneous CSI estimation. The communication channel covering from the base station to RIS and RIS to the users’, and in most circumstances the cascaded CSI of two-hops is appropriate. The cascaded channel gain contains enormous number of channel coefficients because of number of RIS reflecting elements. As a result, a large number of pilot symbols are required for their estimation and, in general, it should be proportional to the number of reflecting elements. As a result, a large number of pilot symbols are required for their estimation and, in general, it should be proportional to the number of reflecting elements. Therefore, the reduction in the channel estimation overhead is an open research problem. In addition, RIS-supported channel estimation does not have any complex signal-processing capability to enhance the problems of the channel estimation. Further, a direct channel between the base station and user could be accomplished by establishing RIS into the absorption mode and then applying the traditional channel estimation approach. Nevertheless, an efficient and advanced approach which maintains power consumption and intricacy of the RIS as small as possible is needed to explore the channel estimation activity for the channels between RIS to base station and RIS to user. There are several techniques which could be used for channel estimation between RIS to base station and RIS to user relying on the hardware resources of the RIS. He and Yuan [29] have presented a cascaded channel estimation approach where a one unit cell is turned on always, even though all the other unit cells are turned off and the product of the channels of that specific unit cell of the base station and user is estimated at the base station for transmitting a pilot signal from the user. Though this technique imposes a long estimation delay because of massive number of RIS elements. Further, this degrades the channel estimation accuracy because the unit cell is always turned ON/active. Moreover, Ning et al. [30] have introduced another estimation technique based on beam training in which the RIS swiftly sweeps the reflection coefficients of its unit cell over a pre-defined codebook instead of clearly estimating the RIS to base station and RIS to user channels. Then, the best beam is selected for the RIS configuration relying on the response of the user’s received signal power. In addition, an alternative approach is exploited by Yuan et al. [31], who have reported the RIS with some active unit cells connected to receive RF chains, and the channel estimation at the RIS is feasible depending on the training signals from the baseband. With considerably high channel correlations between adjacent unit cells in the RIS, precise estimation with a restricted number of active elements could be accomplished at the cost of its complexity and power consumption.

4.2 RIS reconfiguration

The reconfiguration of transceiver in the conventional wireless communication channel has been widely explored. In the RIS-supported wireless environment, the wave propagation channel itself is an optimization parameter and the RIS usually consists of an enormous number of components. Hence, the reconfiguration of RIS in wireless communication use cases would be extremely nontrivial and it has multiple nonconvex constraints which also enhance the complexity of the problem. Nevertheless, the RIS reconfiguration challenges are usually solved by swapping the optimization that could be computationally expensive. Thus, simple algorithms to reconfigure the wireless environment in real-time scenarios are required.

4.3 Network optimization

The real-time configuration of the entire network is a challenging research problem particularly, when it consists of multiple base stations and is supported with various distributed RIS to carry out enormous users. In this scenario, to yield a global optimum over the entire network consist with multiple distributed components and demand huge number of control signals for RISs configuration and resource allocation especially, the power allocation and users’ scheduling.

4.4 Passive beamforming

With this context, we want to estimate the direct channel from the base station to the user by turning off the RIS first, and then estimate the overall channel by turning on the RIS. The difference between the overall channel response and the direct base station–user channel response could be used to compute the cascaded CSI. Since the direct channel from the base station to the user could not be precisely anticipated, the two-channel response subtraction process will further taint the cascaded CSI and cause the error propagation of sequential interference cancellation. Therefore, the cascaded CSI error is important and needs to be taken into consideration. We also have to deal with realistic transceiver hardware faults (HWIs), which are brought on by faulty oscillators, low-resolution “analogue-to-digital converters”, and nonlinear amplifiers. A small number of quantized phase shifts could be used to reduce hardware costs and power consumption, notwithstanding the quantization noise that would be imposed on the phase shifts of RIS unit cells. Due to the presence of “quantized phase noise” at the RIS as contrasted to systems running without RIS, the impact of HWIs on RIS-assisted communication systems is particularly complex. Thus, a vigorous communication design is required considering the HWIs at both the transmitter–receivers and RIS.

The communication networks with RIS are fully able to manipulate the propagation networks by significant variations in the phase and/or amplitude of the incident signal coming from the base station by reflecting its unit cells programmatically and providing passive beamforming by optimizing the reflection coefficients on each reflecting unit cells. Various reported literature studies on RIS-supported wireless communication are restricted to only one-way passive beamforming design, which reveals that phase shift matrices or reflection coefficients from the reflecting unit cells are optimized for either uplink or downlink communication scenarios but not both simultaneously. With appropriate design of phase shifts and attenuations of the RIS reflecting unit cells, the incoming signal/incident signal from the base station reflected by the reflecting unit cells could be superimposed constructively or destructively to the proposed user to improve the attainable link rate performance or overwhelm the interference. Thus, the RIS is able to provide high passive beamforming gain without consuming additional energy for signal regeneration because of its passive characteristics that are different from the traditional “amplify-and-forward” relay.

4.5 RIS-supported frequency-division duplex (FDD) system design

The channel estimation for time division duplex (TDD)-based executions with RIS is a very exciting contribution. However, Yuan et al. [31] reported that the RIS phase shift channel model relies on the angle of incidence, which indicates that the consideration of channel reciprocity in TDD systems would not be held in preparation. So, it is authoritative to explore the channel estimation and channel model for RIS system with FDD. Further, the large-scale channel matrices need to be fed back to the base station with RIS-supported FDD structures due to the huge number of reflecting unit cells in RIS that experience high feedback overhead.

4.6 RIS in terahertz communications

The THz communication offers more plentiful bandwidth and higher data rates as compared to that of the mmWave. However, the THz frequencies suffer from extremely severe path loss, molecular absorption, and strong atmospheric attenuation, which limits its range of operation. In addition, the THz signals are incompetent to support reliable communication links because it is highly prone to blockage effects which limit its practical execution. Therefore, RIS makes THz communication very encouraging because it has an ability to create alternative channel paths. At THz frequency, RIS may hold a massive number of reflecting unit cells which allow us to get a holographical array having a near-continuous aperture. Consequently, beamforming/beam steering and channel estimation for THz communications supported by RIS are an exciting future research direction.

4.7 Mobility management

Mobility management with RIS of the rapid movements of a user is a challenging research problem because the base station might lose its connections with users without using a responsive mobility management scheme. As the RIS is a passive component, it is unable to send pilot signals to track the mobility of the users. Therefore, tracking roaming users is a very challenging task, especially when the direct channel between the base station and mobile users is prevented. Traditional channel estimation methods such as minimum mean square error and least square could be applied to estimate the channel with massive computational complexity. In addition to these methods, the compress sensing-centered channel estimation generally used for leveraging the sparsity characteristics and the channel matrix could also be considered as a noise-centered figure and can be denoised by using deep-learning approach. Further, it is well illustrated that the hybrid neural networks and nonlinear continuous output could be used to enhance the estimation accuracy in high-speed vehicular communication environments.

4.8 Deployment issues

The placement scheme for the RIS unit cells plays a very important role in establishing RIS-related network coefficient and its impact on the performance of the communication system. With context to the practical implementation, the price of hardware, availability of location, distribution of user, and other demanded services, the RIS placements have to be taken into account. In addition to this, the RIS deployment such as centralized and distributed model is a major concern. A centralized channel model in RIS-supported multicell network development is used. In general, the proposed algorithm requires a centralized processing unit (CPU) to collect all complex valued channel matrix over the networks. The weight and phase shift in all active beamforming sent back to the subsequent nodes are computed by the CPU. Centralized algorithms suffer from heavy feedback overhead and high computational complexity, which are major research problems. As compared to conventional (without RIS) structures, the cascaded channel matrix must be fed back to the CPU. Thus, the design of distributed algorithm is very important where the base station could make transmission decisions based on its local CSI with limited information exchange with other base stations. Further, the distributed algorithms have very attractive improvements over the centralized algorithms, diminished computational intricacy, and enhanced scalability.

4.9 Optimization of phase shifts

The IRS-supported wireless networks require the optimization of IRS phase shifts to yield potential objectives for every application. However, the challenge is to optimize the IRS phase shifts efficiently considering the hardware inconsistencies. In general, the phase shift optimization of the IRS-assisted cellular wireless networks is used to increase the throughput of multiuser cellular wireless networks [32]. The phase-shift design for contemporary communication system in a real-world scenario would be more complex than that for the conventional communication systems because the contemporary communication system with RIS would provide smart services to many users at any time.

Currently, artificial intelligence/machine learning-based phase-shift optimization method has been developed as an encouraging route for solving resource distribution challenges. However, an imperfect CSI is unavoidable because of the enormous number of passive reflective unit cells in RIS. In addition, the random error of CSI makes the constraint probabilistic; therefore, the objective function takes extra anticipation. However, to transform the probabilistic constraints that is lower bound on the probability function is equal to the probability of the fulfillment of a certain linear inequality into the deterministic form, and statistical information could be used to compute the probability of the objective function precisely if the distribution of the channels' uncertainty is known. The statistical information of CSI is further made complicated by cascaded channels formed by RIS, and the conversion of stochastic difficulties into deterministic ones are hampered by the performance loss and persistent probability computation. In these circumstances, the approach enabled by “Monte-Carlo simulation” could be used to deal with the channel uncertainty. The response of weighted sum rate (WSR) of users with path-loss exponent decreases with increase of path-loss exponent and converge at no-RIS. The WSR at significantly smaller values of path-loss exponent is very high in case of RIS-based communication systems as compared to that of the random phase as well as no-RIS cases. The mobile users WSR could be optimized by inventing jointly the “active beamforming” at access point and “passive beamforming” at RIS. The performance of WSR is presented in Figure 3 for RIS-related, random phase, and no-RIS, and reveals that the WSR accomplished by the RIS-supported algorithm shrinks with the upsurge of path loss because of the increase in path loss of RIS, and henceforth the signal mirrored by the RIS would be very low. In addition, the WSR is also dependent on the number of reflecting unit cells in the RIS. The generated beam of the signal will be narrow and its information-carrying capacity increases with the increase in reflecting unit cells of RIS as supported by Figure 3. Further, the nature of the presented results is also supported by Pan et al.’s work [33]. This study supports important design intuitions suggesting that the position of RIS should be meticulously selected to avoid obstructions in both the base station to RIS channel link as well as RIS and user channel link.

Figure 3

The response of path-loss model of the RIS on WSR.

The RIS-supported wireless communication networks are transmitter–receiver reciprocal and rely on the consideration that the reflection coefficient is not responsive to variations in the angle of incidence. However, if the phase reflecting unit cells of RIS is responsive to the incidence angle, then channel reciprocity of RIS-supported communications networks cannot stand any further level. The response of angle of incidence in association between the phase shift and control voltage of the RIS unit cells is presented in Figure 4, which reveals that the used RIS is highly susceptible to the incidence angle. Thus, to ensure channel reciprocity, we have developed an RIS that must be insensitive to the incidence angle. The presented results are also supported by Tang et al. [34].

Figure 4

The control voltage response over the reflection coefficient (in degree) of the unit cell to accomplish a relationship with angle of incidence for RIS.