1. Introduction

Visible light communication (VLC), as an optical wireless communication technology using lightwave as the information carrier, has the advantages of wide spectrum, large capacity, wide coverage, high security and low energy consumption [1,2]. It has been identified as an important component in the sixth generation (6G) mobile communication blueprint [3]. Due to its directionality and short-range travel distances of light signals, VLC plays a key role in various indoor interconnection scenarios, such as high-definition video transmission, heterogeneous network, smart home and so on. Meanwhile, it is well known that the acquisition of high-precision indoor positioning information is the foundation of intelligent equipments and Internet of Things (IoT) [4], which directly affects the development of smart manufacturing, smart medical care, and next-generation robots. It is also one of the urgent problems to be solved in 6G. Depending on existing lighting devices, VLC can also be applied to some scenarios that require high-precision positioning information, and visible light positioning (VLP) technology has emerged [5]. Compared with traditional positioning technology, VLP can not only enable higher accuracy ranging from several to tens of centimeters, but also provide lower system complexity and paving cost [6,7]. A tested VLP system based on LED arrays supporting 15 independent channels was presented in [8], which can enable mobile users obtain higher positioning accuracy. However, most of the current researches on VLC and VLP are independent, without considering that VLC is the basis for obtaining light intensity information of VLP [9]. To address this issue, a new integrated visible light communication and positioning (VLCP) system based on filter bank multicarrier-based subcarrier multiplexing was firstly proposed and demonstrated for realizing both high-speed communication and high-accuracy positioning services [10]. In [11], three conceptual designs were presented for VLCP system, which can effectively utilize channel and location information to improve communication and positioning performance. However, these works only focus on single user scenario. A multi-layer network architecture with effective optimization was proposed for integrated VLCP system supporting multiple IoT users, which verified the superiority in the performance of integrated VLCP IoT network [12,13].

The ability to support multiple users in the VLCP network is a great challenge for practical applications. Both VLC and VLP systems adopt multiple access technologies to eliminate or reduce multiple access interference (MAI) [14]. An integrated VLCP system with orthogonal frequency division multiple access (OFDMA) was presented in [15]. However, the system has higher peak to average power ratio (PAPR) and implementation complexity, owing to its multi-carrier modulation format used in transmission signal. In [16], the multi-band carrierless amplitude and phase (m-CAP) technology with lower PAPR was investigated for VLCP system for the first time, reducing system bit error rate (BER) and improving positioning accuracy. A filter-enhanced VLCP scheme was presented and evaluated in [17], where spatial waveform shaping filters are applied to enable the received signals to be separated, but system implementation complexity was increased. The code division multiple access (CDMA) technology with single carrier modulation can be viewed as one efficient and straight forward way to reduce or eliminate MAI, which has lower PAPR and implementation complexity [18–20]. It is worth noting that most of VLP systems only employ CDMA technology to distinguish identification (ID) and obtain light intensity information of LEDs, so as to calculate the accurate positioning information [21,22]. Actually, CDMA technology with different spreading code sets can also recover the original user data effectively, which lays the foundation for realizing integrated VLCP system [23]. Adopting CDMA technology in VLCP system can better recover original user data and obtain positioning information simultaneously, which is more efficient and suitable for practical intelligent systems [24].

The design of spreading code set is extremely important in CDMA communication systems [25], which should also be noticed in VLCP-CDMA system according to its communication and positioning demands. Due to the widespread adoption of intensity modulation (IM) in VLCP system, traditional bipolar spreading code sets such as Walsh code set cannot be directly used. A fixed DC offset was utilized to convert the bipolar CDMA signals to unipolar signals [18,26], leading to increased energy consumption. In [19], the transmitter directly set the negative codes ‘−1’ to ‘0’ and kept each ‘1’ unchanged. However, it may introduce higher MAI due to the destruction of original code set orthogonality. Some unipolar code sets have been constructed in [27,28], but the introduction of too many ‘0’s for ensuring correlation properties would result in worse illumination performance. Furthermore, it is difficult to achieve perfect synchronization among users in CDMA system with time delay or multi-path transmission, thus some researches concentrated on quasi-synchronous (QS) system [29–31], of which time delay among user transmission signals is within a certain range. In our previous work [32,33], optical zero correlation zone (OZCZ) code sets have been proposed for QS CDMA-VLC system, which can effectively tolerate inevitable time delay among users by theoretical and experimental analysis. But the number of constructed code sets is limited, which may influence the flexibility of system.

The main contribution of this paper can be summarized as follows:

The paper is organized as follows. After introducing the modification method and correlation properties analysis for bipolar code sets in section 2, we describe the proposed VLCP-CDMA system model design and principle in section 3. Then, we elaborate the system setups in Section 4, together with the simulation and experimental results and discussions. Finally, we draw the conclusions in Section 5.

2. Modification method for balanced bipolar code sets

2.1 Modification method

Let $\boldsymbol {C}=\left \{\boldsymbol {c}_i\right \}_{i=1}^{K}(c_{i,j}\in \{-1,1\},0\leq j<L)$ denote a bipolar code set with $K$ codes, each has length $L$. The number of ‘1’s in each code of the code set is defined as weight $w$. When the code set’s weight satisfies $w=L/2$, it can be called balanced bipolar code set, which will be discussed in the following analysis. The periodic cross-correlation function (PCCF) between each balanced code at shifts $\tau$ can be defined as follows:

If the absolute values of periodic correlation functions for the code set satisfy:

From Eq. (4), we can see that the modification method would not affect the correlation properties, which enables different users in VLCP-CDMA systems can still obtain the desired information effectively.

2.2 Examples of modified Walsh and ZCZ code set

The modification method can be applied to balanced bipolar code set, such as balanced Walsh and balanced ZCZ code set with zero correlation properties, which we would modify and adopt as spreading code set in the following proposed VLCP-CDMA system for comparison.

1) Modified Walsh code set

For constructing balanced Walsh code set, the Hadamard matrix $\boldsymbol {H}_r$ should first be generated as follows

2) Modified ZCZ code set

Due to time delay and multi-path transmission among different users, non-perfect synchronous problems should also be noted in VLCP systems. The ZCZ code set has ideal auto-correlation and cross-correlation properties in zero correlation zone, which can reduce the MAI in QS VLCP-CDMA system effectively [34]. There are many construction methods for balanced ZCZ code set [35,36]. Based on interleaving and iteration method [32], a balanced bipolar ZCZ code set $\boldsymbol {Z}=\left \{\boldsymbol {z}_i\right \}_{i=1}^{K}$ can be obtained by $r$-iterations of the initial matrix $\boldsymbol {I}_0$

Adopting modification method of Eq. (3), the unipolar code set $\boldsymbol {Z}'$ corresponding different user data $u_k$ can also be obtained, and a modified ZCZ code set pair $<\boldsymbol {Z},\boldsymbol {Z}'>$ is constructed. Assuming that $Z_{cz}$ is zero correlation zone length, the parameters of this code set pair satisfy $(L,K,Z_{cz})=(4KZ_{cz},K,Z_{cz})$, while the absolute values of PACF and PCCF follow as

Thus, compared with the modified Walsh code set, the modified ZCZ code set has cross-correlation functions of zeros within $Z_{cz}$, in other words, it provides lower side-lobes, making it easier to reduce MAI among users [37].

3. System design and principle

Figure 1 gives the proposed VLCP system design with CDMA technology, which is used in the subsequent performance analysis. For simplicity, we only focus on the line-of-sight (LOS) case. In the transmitter, the signal control unit is introduced to load different spreading signals to LEDs in parallel. Each LED is assigned a unique spreading code corresponding to each user. To reduce system complexity, different LEDs transmit on-off keying (OOK) signals with single carrier. A reasonable layout of light source LEDs is adopted to enable each user can receive optical signals from more than three LEDs. In the receiver, the user data and positioning information acquisition unit based on a single photo detector (PD) firstly realize photoelectric conversion by I-V circuit, and store the electrical signals in signal acquisition and storage module. The despreading signals are used for communication and positioning respectively. The desired user data is recovered by adopting the unique spreading code and despreading signals to realize decision. The acquisition of positioning information is through the measurement of different despreading signals for determining ID information, estimating channel gains and transmission distances. The principle of VLCP-CDMA system process is presented as follows:

 

Fig. 1. Block diagram of the VLCP-CDMA system.

Download Full Size | PDF

The $i$-th original user data of the $k$-th user can be expressed as $u_{k,i}\in \{0,1\}$, while the $K$-user system employing $<\boldsymbol {C},\boldsymbol {C'}>$ with length $L$ as spreading code set pair. In order to suit IM of VLCP systems, the transmitter and receiver would use unipolar code and bipolar code respectively. In the transmitter, the $l$-th time-domain chip of the $k$-th user transmitting signal after spreading with unipolar code $\boldsymbol {c}'_{k\_u_{k,i}}=\{c'_{k\_u_{k,i},1},c'_{k\_u_{k,i},2},\ldots,c'_{k\_u_{k,i},l}\}$ can be given by

On the basis of a generalized Lambertian radiation model [38], the optical channel gain $H_{k}$ between the $k$-th transmitter and the terminal receiver can be expressed as

Assuming that the average transmitted power of LED is set as $P_t$, the received signal $r_{i,l}$ corresponding the $l$-th chip of the $i$-th user data from $k$-th LED can be represented as

In the receiver, the spreading code $\boldsymbol {c}_d=\{c_{d,1},c_{d,2},\ldots,c_{d,l}\}$ corresponding to the $d$-th user is used to despread as follows

Combining Eq. (3), Eq. (9) and Eq. (14), we have

From Eq. (15), we conclude that the MAI from other users can be removed by spreading and the amplitude of the signal is increased by $L$ times. With the same number of ‘$+1$’ and ‘$-1$’ in each code, the term $\sum _{l=1}^{L}\sum _{k=1,k\ne d}^{K}\Re P_tH_kc_{d,l}/2$ is zero. The term $\sum _{l=1}^{L}\sum _{k=1,k\ne d}^{K}c_{d,l}n_{k,i,l}$ is much less than $L\Re P_{t}H_{d}(-1)^{u_{d,i}}/{2}$. Therefore, the signal $S_{d,j}$ is approximately equal to $L\Re P_{t}H_{d}(-1)^{u_{d,i}}/{2}$, which lays the foundation of recovering the original user data and obtaining the positioning information in VLCP-CDMA systems, respectively. At the same time, lower side-lobes of correlation functions would also reduce MAI during optical signal transmissions, which can enhance the system communication and positioning performance [34,40]. When considering time delay and multi-path of QS systems, zero correlation zone properties of code sets are helpful for reducing the effect of time delay among different users and guaranteeing the system performance [32].

1) Obtaining the positioning information

Received signal strength (RSS) positioning algorithm are widely used in VLP system, due to its high-accuracy and low-complexity [41,42]. For simplify, we utilize the two-dimensional (2D) trilateral positioning algorithm based on RSS, which adopts three stronger received signals to calculate the positioning information. Other RSS positioning algorithms with higher positioning accuracy and three-dimensional (3D) positioning function can also be applied to the proposed system [43]. The process of obtaining the positioning information is given as follows:

Firstly, the $j$-th $(j=1,2,3,4)$ channel gain corresponding to RSS information can be calculated by

Then, it is assumed that the position coordinates of the three LEDs are $(x_1,y_1,z_1)$, $(x_2,y_2,z_2)$, $(x_3,y_3,z_3)$, respectively. The terminal receiver is located at $P=(x,y,z)$. When considering that the distance between the transmitting plane and the receiving plane is the same as the height of space $h$, the distance from different LEDs to the receiver can be expressed as

For the specific system, the parameters of LEDs, PD and system are fixed, Eq. (18) can be simplified as

The coordinate of $P$ can be obtained by

2) Recovering the original user data

For communication, the original user data needs to be recovered by despreading and decision. If the decision threshold is $Th$, the original user data can be recovered by

The appropriate threshold can be selected according to the polarity of the spreading code. For the system with unipolar code set for despreading, the decision threshold $Th$ needs to be dynamically adjusted based on the range of the received signal. For the system with the modified bipolar code set pair, the signals after despreading become bipolar signals, therefore the decision threshold $Th$ is always 0, which can reduce the system complexity effectively.

4. System setup, results, and discussions

We conduct the proposed VLCP-CDMA system with 4 LEDs for simulation and experiment, employing the modified ZCZ code set and modified Walsh code set with the same code length $L=16$ and weight $w=8$. The number of LEDs can be increased according to the illumination, communication and positioning requirements of users in the indoor space, some works have concentrated on layout of LED light source and allocation of code sets and optical cell [44,45]. In this work, we assume that each LED is assigned a unique spreading code, and each user can receive signals from all four LEDs at the same time. The terminal receiver is horizontally placed on the receiving plane, $11 \times 11$ sampled points are evenly placed and evaluated in the simulation and experiment. Each samples points are measured 10 times. The data rate of each user can be obtained by $R=S/L$, where $S$ is the transmission rate of LED transmitted signals [33].

4.1 Simulation system setup and results

In the simulation, the space size is set as 6 m $\times$ 6 m $\times$ 5 m. The four LEDs are evenly distributed on the horizontal ceiling, and the coordinates of their positions are $(1.5, 1.5, 0)$, $(4.5, 1.5, 0)$, $(4.5, 4.5, 0)$ and $(1.5, 4.5, 0)$ (unit: m), respectively. The rest simulation parameters are shown in Table 1.

Table 1. Simulation parameters of the VLCP-CDMA system.

View Table

Figure 2 compares the simulation results of BER and average positioning error performance versus transmitted power of each LED in the synchronous VLCP-CDMA system adopting modified ZCZ code set and modified Walsh code set. The data rate of each user $R$ is set as 1 Mbit/s, 2 Mbit/s and 3 Mbit/s, respectively. The average positioning error is defined as the average of all users positioning errors at 121 sampled points in the following analysis. The results in Fig. 2(a) show that with the increased transmitted power of each LED, the system BER performance can be significantly improved with the increased signal-to-noise ratio (SNR). Besides, due to the influence of high-frequency fading characteristics of LEDs, the transmitted signals with the higher transmitted rate have lower received power, leading to the decrease of system BER performance. In Fig. 2(b), it is demonstrated that, the communication and positioning both depend on despreading, thus with the higher transmitted power and lower transmitted rate, the average positioning error of the system would be reduced as well. When the received SNR gets large enough, the difference between the acquired light intensity information is small, the average positioning error will be essentially unchanged. Furthermore, for the reason that the modified ZCZ code set has lower side-lobes correlation properties than the modified Walsh code set, the modified ZCZ code set can reduce system BER and enhance positioning accuracy more effectively, which can be applied to the system with lower transmitted power and higher data rate.

 

Fig. 2. Simulation results of (a) system BER and (b) average positioning error performance versus the transmitted power of LEDs when the the data rate of each user is 1 Mbit/s, 2 Mbit/s and 3 Mbit/s.

Download Full Size | PDF

The simulation results of system BER and average positioning error performance versus the time delay using modified ZCZ code set and modified Walsh code set in VLCP-CDMA system are showed and evaluated in Fig. 3. The transmitted power of LED and transmission rate in the simulation are set as 10 W and 10 Mbit/s, respectively. It is well known that time delay among users may be introduced by the difference of transmission distance, multipath effects and so on. When the time delay is 0 $\mu$s, the system is synchronous system. When the time delay is greater than 0 and less than 1 $\mu$s, i.e., when the delay does not exceed the zero correlation zone length of adopted code set, the system is considered as QS system. In the simulation, we consider the case of time delay between the first user and other users for example, similar simulation results will be obtained when there is time delay between other users as long as the time delay is guaranteed to be within the zero correlation zone. It can be clearly observed from Fig. 3, when considering time delay, the VLCP-CDMA system using modified ZCZ code set has obviously better communication and positioning performance than using modified Walsh code set, and the system BER and average positioning error with the modified ZCZ code set fluctuate very little as the time delay increases. The reason is that the ideal zero correlation properties of the modified ZCZ code set can effectively overcome small time delay among users and reduce the effect of MAI. At the same time, the longer zero correlation zone length would tolerate the longer time delay [32,33]. It is necessary to make a trade-off between transmission efficiency rate and BER performance according to system requirements.

 

Fig. 3. The simulation results of (a) system BER and (b) average positioning error performance versus the time delay using different code sets in the VLCP-CDMA systems.

Download Full Size | PDF

4.2 Experimental system setup and results

As shown in Fig. 4(a), an integrated VLCP-CDMA real-time system with 4 LEDs is realized in a indoor space of 60 cm $\times$ 60 cm $\times$ 50 cm. The transmission rate of transmitted LED signal and the sample rate is set as 1 Mbit/s and 5 MS/s, respectively. For ensuring that all points in the space can receive signals from at least three LEDs, four 1W OSRAM LEDs are used in the experiment, and its layouts are presented in Fig. 4(b). The original user data is stored in signal control unit realized by field programmable gate array (FPGA), and then loaded on each LED by drive circuit (BUF 634). A PD (Hamamatsu S6801) with 1 $cm^2$ detection area is used for the I-V circuit to receive the optical signal and realize photoelectric conversion. An ARM and FPGA integrated hardware architecture (Winner I) is adopted to realize the user data and positioning information acquisition unit. At the same time, for suiting real experimental system, a fitting method is introduced to obtain related positioning parameters and estimate different distance [41].

 

Fig. 4. (a) The 4 LEDs VLCP-CDMA real-time experimental system, (b) LEDs layout.

Download Full Size | PDF

Figure 5 presents the experimental CDF results of BER and positioning error performance with different code sets in synchronous and QS VLCP-CDMA systems. One codeword delay for the first user is also introduced in the experimental QS VLCP-CDMA system. As we can see from Fig. 5, the communication and positioning performance results show a similar tendency with those in Fig. 3. Due to its lower side-lobes of correlation functions than the modified Walsh code set, the modified ZCZ code set can reduce the average positioning error by $34.21\%$ in the synchronous VLCP-CDMA system. When considering the QS system with time delay, the average BER performance using modified ZCZ code set is $95.72\%$ lower, while the average positioning error is $69.30\%$ lower. The results demonstrate that the modified ZCZ code set with ideal correlation properties in the zero correlation zone performs better in the experiment considering multipath, difference of transmission distances and other effects. Besides, limited by experimental system condition, the BER performance of the measured points in the QS system using modified ZCZ code set is only $37.19\%$ under the FEC threshold limit of $3.8\times 10^{-3}$, but it is all above the FEC threshold limit when using the modified Walsh code set. The further improvement of system performance can be realized by adopting new devices and channel equalization technology.

 

Fig. 5. The experimental CDF results of the (a) system BER and (b) positioning error performance with different code sets in the synchronous and QS VLCP-CDMA systems.

Download Full Size | PDF

The BER distributions of the experimental VLCP-CDMA system using modified ZCZ code set and modified Walsh code set are presented in Fig. 6(a) and (c), which is similar to the simulation result. Due to the Lambertian radiation model, the SNR of the received signal at the edge of space is lower than that at the center leading to a degradation of BER performance at the edge of space. The $85.12\%$ sampled points are below the FEC threshold limit, and the average system BERs are $1.8\times 10^{-3}$, when using different modified bipolar code sets. Figure 6(b) and (d) show the positioning distribution of measured points and sampled points in the system. The blue squares represent 121 sampled points, and the red dots represent the corresponding measured points. It can be seen that the measured points are distributed near the sampled points, and positioning error at the edge of space is greater than the centering positioning error, which is similar with the system BER distribution and satisfies the position error distribution feature of trilateral positioning algorithm. The maximum, minimum and average positioning errors of the system with modified ZCZ code set achieve 5.50 cm, 0.26 cm, and 1.50 cm, respectively, while the maximum, minimum and average positioning errors using modified Walsh code set are 7.93 cm, 0.3 cm and 2.28 cm, respectively.

 

Fig. 6. The BER distributions of the experimental VLCP-CDMA system with (a) modified ZCZ code set and (c) modified Walsh code set, and the positioning distributions of measured points and sampled points in the system with (b) modified ZCZ code set and (d) modified Walsh code set.

Download Full Size | PDF

5. Conclusions

In this paper, we have proposed and realized an efficient integrated VLCP-CDMA system for the requirements of communication and positioning in 6G. A generalized modification method is used for balanced bipolar code sets, allowing them to be suitable for the proposed system. The simulation and experimental results verify that the modified ZCZ code set with zero correlation zone properties can be adopted to improve the BER and average positioning error performance in VLCP-CDMA system more effectively, even in the QS system with a small time delay among users. A real-time VLCP-CDMA system with the BER of $1.8\times 10^{-3}$ and average positioning error of 1.50 cm are successfully achieved in a space of 60 cm $\times$ 60 cm $\times$ 50 cm. Therefore, it is revealed that the modified ZCZ code set can be considered as a potential and suitable candidate for the proposed VLCP-CDMA system, which can significantly overcome non-perfect synchronous problem.