1.1 Motivation

With the surge in wireless devices connected to wireless networks, the radio spectrum resources are becoming limited. With the rise of data intensive applications like online gaming, video streaming, high definition (HD) multimedia streaming with services like Netflix and Youtube, large data sets need to be transmitted and processed [1]. The proliferation of smart sensors, Internet-of-things (IoT) devices, smart phones, smart wearables also accumulate massive data to be transmitted in real-time [2]. This growing amount of data traffic frequently exceeds the capacity of the available spectrum leading to access congestion and a decline in service quality [3]. Moreover, there is a significant upswing in the energy consumption owing to this rapid proliferation [4]. In conventional wireless communication, radio frequency (RF) is used to access the wireless networks. However, to support uninterrupted streaming and real-time sensor data exchange in IoT with reduced energy overhead, novel energy efficient communication solutions and spectrum resources are required [5,6]. Visible light communication (VLC) is emerging as a promising solution. It has huge potential to support the high demand of data traffic and offer huge energy savings [7]. VLC utilises the visible light spectrum to transmit the information using optical signals [8]. Since it serves the dual purpose of illumination and communication with the same power, the power consumption is relatively low [9]. The existing lighting infrastructure have LEDs that act as VLC transmitters. The photodiodes (PDs) are used at the receiver end for VLC signals reception. However, due to its reliance on line-of-sight (LoS) propagation links between the transmitter and receiver, it often results in signal blockages and limited coverage [10]. In this case, VLC-RF coexisting systems are a potential candidate for future communication systems [11]. While RF can offer wide coverage and reduce VLC inter-cell interference, VLC uses the spectrum resources in the visible light band to reduce spectrum congestion of RF communications and enhance energy efficiency.

1.2 Related work

The literature includes number of research works on VLC-RF systems. However, there are two variations of VLC-RF systems, namely hybrid VLC-RF systems [12] and aggregated VLC-RF systems [13]. The users can be simultaneously served by VLC access point (AP) and RF AP in case of aggregated VLC-RF system. Whereas in case of hybrid VLC-RF systems, the user is either served by VLC AP or RF AP. Due to the presence of heterogeneous APs in such communications systems, they suffer from the challenges of AP assignment, efficient resource allocation and power control [14]. There are number of research papers that evaluate the performance of VLC-RF networks. For example, [15] investigates an AP assignment problem jointly with power allocation in hybrid VLC-RF network for maximum sum rate in a multi-tier heterogeneous network. The authors in [16] consider the power allocation problem in aggregated VLC-RF system. Based on orthogonal frequency division multiple access (OFDMA) the energy efficiency of the system is maximised with optimal subchannel allocation. For the VLC-RF based wireless systems, the important communication metrics are data rates, spectral efficiency and energy efficiency. To achieve the optimal communication metrics, the signal processing techniques like resource allocation, power allocation, AP assignment need to be considered.

Since the VLC system performance is dependent on the LoS link quality, reconfigurable intelligent surfaces (RISs) have been introduced in VLC systems [17]. RISs are intelligent surfaces or metasurfaces that control the reflection of incoming waves enabling smart radio [18]. These are also called intelligent reflecting surfaces (IRSs). By dynamically controlling the phase shifts of number of RIS reflecting elements, directed beamforming is enabled offering LoS communication links. The RIS controller adjusts the phase shifts so that the wireless channel can be altered. In order to increase the signal strength at the desired receiver, there is constructive interference of reflected signals while signals interfere destructively in other directions [19]. These consume less power owing to no requirement of RF transceiver chains and power amplifiers. Apart from passive IRSs, active IRSs have also been introduced that allow selective amplification of reflected signals [20]. Another variation of IRSs is the STAR-RISs which is simultaneous transmission and reflection RISs. These enable selective reflection to the users in the reflection zone as well as selective transmission to the nodes in the transmission zone simultaneously [21]. The use of RISs in VLC-RF systems have been gaining interest. These enhance the communication performance and improve the energy efficiency. In the RF systems, a number of works focus on the assistance of IRSs for improved performance. The authors in [22] consider an IRS aided MIMO system for maximum sum rate. Through the optimization of base station (BS) precoding and RIS phase shifts, channel overhead is reduced using statistical channel state information (CSI). The IRS technology finds usage in diverse applications like mobile edge computing [23], network security [24] and wireless power transfer (WPT) [25]. The assistance of IRSs in the VLC system is elaborated in [26] highlighting the role of RIS technology in addressing the skip-zone problem. The authors in [27] maximize the spectral efficiency in the IRS assisted VLC environment through joint resource utilisation. For focussing the optical power towards a VLC receiver, [28] proposes two types of IRSs based on mirrors and programmable metasurfaces. The orientations of the arrays are adjusted in order to achieve maximum received power gain. The delay in signal transmission from source to the receiver via IRS is calculated in [29] for an IRS-based VLC system. The channel characteristics for the IRS VLC system model are also studied. [30] uses optical IRSs to overcome the challenge of spatial correlation in VLC channels leading to low spatial multiplexing gain. To obtain tradeoff between performance and complexity, the optical IRS are optimised using two algorithms for optimal RIS element alignment and source emission power. The potential of IRSs in free space optical (FSO) systems is highlighted in [31] for high rate secure communication. The comparison of optical IRSs with RF IRSs is also presented.

The performance of hybrid VLC-RF system is investigated in [32] for maximum energy efficiency and sum achievable rate. The allocation of optimal power with optimal association of transceivers optimization yields quality-of-service (QoS) to the multiple users in the system. The system energy efficiency, sum rate and outage performance are evaluated in [33] for aggregated VLC-RF networks through joint power allocation, AP assignment and subchannel allocation. It outperforms the hybrid VLC-RF system and achieves considerable system gains. The performance of aggregated VLC-RF system is elaborated in [13] for maximum energy efficiency subject to transmit power constraints and dimming control. As compared to conventional RF system, the considered system achieves improved data rate. Very few papers have introduced IRSs into the hybrid VLC-RF systems.

In this paper, an integrated VLC-RF system is considered in which multiple ORISs assist the VLC communication while an RF RIS assist the RF communication. The use of RISs dynamically alters the propagation of signals enabling smart radio. The main contributions of the paper are summarised as

• An integrated VLC-RF system is considered to avail the advantages of both VLC and RF communication. VLC offers broad spectrum and high speed communication.
• To mitigate the challenge of finite coverage and signal blockages in VLC as well as effects of wireless channels in RF, ORISs and RF RIS are deployed in the system scenario. These dynamically adjust the signal propagation to enhance the system performance.
• For better utilisation of resources, an association algorithm is proposed which allocates a set of ORISs to each VLC AP-node pair for maximum system sum rate.
• The hybrid VLC-RF system is evaluated for system sum rate and energy efficiency. A power consumption of the hybrid system is also obtained.
• In the end, the impact of power allocation on the system performance is carried out. The performance comparison with the conventional VLC system and RF system is also investigated.

2.1 VLC channel modelling

Suppose the user nodes being served by VLC AP v is denoted by a set , with denoting the complete set of VLC APs. To model the VLC communication between node k and VLC AP v, Lambertian channel model is used as follows

(1)

with G denoting the optical filter gain, being the refractive index, is the semi angle of PD FoV. is the Lambertian index, is the distance between node k and VLC AP v, A_k is the PD area on receiver k, is the node k PD’s incidence angle with is the source LED array emission angle on VLC AP v.

The association between node k and VLC AP v is defined by an association matrix defined as

(2)

where denotes the closest VLC AP to node k such that .

To represent the ORIS reflected channel path, VLC AP v transmission to node k is assisted by element of ORIS as follows

(3)

with being the ORIS reflectivity, is the source emission angle on ORIS unit k. The incidence angle on the node k from ORIS unit n is denoted by while r_k,n and be the distance between node k to ORIS element n and distance between ORIS element n to VLC AP v respectively. The overall channel gain in the VLC system is given by

(4)

where is an assignment variable whose value is 1 when ORIS element is assigned to serve node k. Its value is zero otherwise.

2.2 RF channel modelling

In order to model the direct RF channels between RF AP and the nodes, Rayleigh fading channel is used. Denoting the path loss by at reference distance of 1 m, for the RF AP to node k link with distance r_k, the direct channel from RF AP to all the nodes is given by

(5)

with with representing the Gaussian distributed elements which are independent and identically distributed with zero mean and unit variance.

Since the channels from RF AP to the IRS and IRS to the nodes are LoS channels, so Rician fading model is used to characterise these channels. The channel from RF AP to the element of RIS is given as follows

(6)

with and being the path loss and distance between RF AP and RF RIS. On similar lines, the channel from RF RIS to the node k is

(7)

with and being the path loss and distance between RF RIS to the node. and follow Gaussian distribution. The channel characteristics from RF AP to RF RIS and that of RF RIS to all the nodes is represented by

(8)

(9)

2.3 RF RIS properties

The properties of RF RIS is represented by a reflection matrix where represent the reflection coefficients of the N elements of RF RIS. Each element reflection coefficient is denoted by with amplitude coefficient A_n and phase shift . Thus, and and . The total channel gain is the sum of direct channel and RF RIS reflected channel given as

(10)

2.4 System achievable rate

The system achievable rate is the sum of achievable rate of VLC communication system and RF communication system given as

(11)

where

(12)

(13)

with being the SINR given by

(14)

where is the power allocated to VLC AP v, B₁ is the VLC communication bandwidth and is the noise power spectral density (PSD).

(15)

where , is the power received at node such that the power vector is represented as , is the noise PSD and B₂ is the bandwidth of RF communication.

To carry out the energy efficiency of the system, it is required to obtain the total power consumption as defined below.

(16)

The total power consumption P_tot is the sum of hardware power consumption P_H and transmit power consumption P_Tx.

(17)

(18)

where P₁ is the power consumed by VLC AP, P₂ is static power consumption of RF AP, P₃ denotes the power consumed by each element of RF RIS while P₄ represents the power consumption of ORIS element.

The transmit power consumption P_Tx is given by

(19)

(20)

where Q is the zero forcing precoding matrix at RF AP.

(21)

3.1 Proposed association algorithm

Let us suppose denotes the user groups assigned to the ORISs . Initially each user node k in the set selects its nearest VLC AP v. This nearest AP will serve as the master AP for node k as it offers strong channels for v–k transmission. It is to ensure that maximum number of user nodes served by each VLC AP and each ORIS should not exceed the defined threshold , that is, length and length . After that, for each v–k pair transmission, the SINR at node k is calculated. The RF SINR owing to RF communication at node k is also obtained. The optimal ORIS that maximises the total SINR at node k is selected and assigned for particular v–k pair. The user set is updated with the node k. This is repeated for all VLC AP and node pairs.

The Algorithm 1 explains the proposed association approach. The flow of operations associated with the proposed algorithm is summarised in the flowchart shown in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Illustration of the flow of operations in the proposed algorithm.

https://doi.org/10.1371/journal.pone.0327467.g002

Algorithm 1. Optimal ORIS association algorithm.

Input: , , , , , , , , , B₁, B₂,

Output:

  1. Initialization

  , , ,

  2. Master VLC AP association

  3. Find , for VLC communication for each (–) pair

  4. Find for RF communication

  5. Find total SINR at each node

  6. For each (–) pair, select optimal ORIS l^*

  7. Associate that node to the user set served by optimal ORIS

  l^*

  8. Repeat for all the – pairs

  return

The flowchart starts with defining the input and output parameters. The first step is the initialization of the parameters and threshold value. The next step is the master VLC AP association in which each user node k finds its master VLC AP . After that, the SINR at each node is obtained owing to both VLC communication and RF communication . The total SINR at each node is calculated as .The ORIS that maximizes is obtained. In the end, that node is assigned to the subset . The output is returned.