Experimental demonstration of quasi-synchronous CDMA-VLC systems employing a new OZCZ code construction

Danyang Chen; Jianping Wang; Huimin Lu; Lifang Feng; Jianli Jin

doi:10.1364/OE.27.012945

1. Introduction

With the tremendous progress and widespread application of light emitting diodes (LEDs) in illumination, visible light communication (VLC) has become an emerging technology and attracted many researchers’ attentions. It has been considered as a promising complement for fifth-generation (5G) wireless communications for the wider bandwidth, lower power consumption, greater security, more flexibility and no electromagnetic interference [1,2]. The ability to support multiple users in the VLC network is a great challenge for future applications. Code Division Multiple Access (CDMA) technology has been introduced as one efficient and straight forward way to reduce or eliminate multiple access interference (MAI) [3,4]. A novel VLC multiple access scheme using joint color-shift keying (CSK) modulation and CDMA has been presented, with a good bit error rate (BER) performance by using Walsh-Hadamard codes as spreading codes [5]. In [6], a single cell CDMA-VLC system was proposed for electromagnetic-wave-free indoor healthcare service by using 3-level amplitude signals to simplify the system implementation. Resource allocation in a multi-cell CDMA-VLC system has been investigated and discussed in [7,8], based on maximizing data rate and achieving fairness in the network.

Due to the widespread adoption of intensity modulation and direct detection (IM/DD) in VLC systems, the bipolar spreading codes in traditional wireless CDMA cannot be directly applied. A fixed DC offset was utilized to convert the bipolar CDMA signals to unipolar signals in [5,9]. In [10], the transmitter directly set the negative codes ‘-1’ to ‘0’ and kept each ‘1’ unchanged. An improved CDMA scheme, using the mapping module to convert the bipolar signals, has been successfully implemented based on field programmable gate array (FPGA) [11]. However, these modifications of the bipolar codes may introduce higher MAI due to destruction of the original orthogonality of codes. A lot of unipolar codes with good correlation properties have been constructed, including optical orthogonal codes (OOC) [12,13], random optical codes (ROC) [14] and prime codes (PC) [15] to adapt to the CDMA-VLC system. It is known that, due to non-perfect clock or multi-path transmission, perfect synchronization is difficult to be achieved in VLC-CDMA system. Most of the prior works mainly focused on ideal synchronous CDMA system, with an emphasis on the improvement of BER performance. For quasi-synchronous CDMA-VLC (QS-CDMA-VLC) system, the design of optical zero correlation zone (OZCZ) codes [16,17] and zero cross correlation (ZCC) codes [18,19] have been proposed to tolerate the inevitable time delay among users with theoretical and numerical analyses. To ensure the correlation properties, the introduction of too many ‘0’s in the generated unipolar optical codes results in low dimming values, which influences the illumination performance and is another concern in VLC system using lighting source, particularly for the multi-user environment [20,21].

In this paper, we propose a new construction of an OZCZ code set by interleaving and iterating methods. The correlation properties and advantages of the proposed code set are derived and discussed. For the VLC system using the proposed code construction, the transmitter adopts unipolar codes to be more suitable for intensity modulation, while the receiver adopts bipolar codes to get desirable signals effectively. We further introduce the proposed OZCZ codes to a QS-CDMA-VLC system. The BER performance of the VLC system based on the proposed OZCZ codes is investigated theoretically and experimentally for different user numbers, sample rates, delay user numbers and transmission distances. To our best knowledge, this is the first time that the OZCZ code set is evaluated in experimental QS-CDMA-VLC system.

The rest of the paper is organized as follows. Section II introduces the preliminaries. The new construction of OZCZ code set and the derivation of properties are elaborated in Section III, together with a specific example. Section IV presents the system model and experiment setup. The results of simulations and experiments for QS-CDMA-VLC system are addressed in Section V. Finally, Section VI draws the conclusions.

2. Preliminaries

Assuming two codes $x_{i} = [x_{i, 0}, x_{i, 1}, ..., x_{i, L - 1}]$ and $y_{j} = [y_{j, 0}, y_{j, 1}, ..., y_{j, L - 1}]$ with length L, their Periodic Cross-Correlation Function (PCCF) is defined as follows:

θ_{x_{i}, y_{j}} (τ) = \sum_{l = 0}^{L - 1} x_{i, l} y_{j, (l + τ) \mod L} \forall τ \geq 0,

when

x_{i} = y_{j}

, it becomes the Periodic Auto-Correlation Function (PACF).

Definition 1: Let $X = {x_{i}}_{i = 1}^{K} (x_{i, j} \in {0, 1}, 0 \leq j \leq L - 1)$ denote an optical zero correlation zone (OZCZ) code set with $K$ codes, each has length L and zero correlation zone $Z_{c z}$ , and the following correlation properties:

θ_{x_{i}, x_{j}} (τ) = {\begin{array}{l} \begin{matrix} \pm w & i = j, τ = 0 \end{matrix} \\ \begin{matrix} 0 & i \neq j, τ = 0 \end{matrix} \\ \begin{array}{l} 0 & 0 < | τ | \leq Z_{c z} \end{array} \end{array},

where w denotes the weight of the code, which is the number of ‘1’ in the code.

Definition 2: Let $I (\cdot)$ denote the interleaving method. Two generalized codes $u$ and $v$ with length 2L can be constructed based on $x_{i}$ and $y_{j}$ as follows

u = I (x_{i}, x_{j}) = [x_{i, 0}, x_{j, 0}, x_{i, 1}, x_{j, 1}, ..., x_{i, L - 1}, x_{j, L - 1}],

v = I (y_{i}, y_{j}) = [y_{i, 0}, y_{j, 0}, y_{i, 1}, y_{j, 1}, ..., y_{i, L - 1}, y_{j, L - 1}] .

The PCCF between the generalized codes

u

and

v

can be calculated by [22]

θ_{u, v} (2 τ) = θ_{x_{i}, y_{i}} (τ) + θ_{x_{j}, y_{j}} (τ),

θ_{u, v} (2 τ + 1) = θ_{x_{i}, y_{j}} (τ) + θ_{x_{j}, y_{i}} (τ + 1) .

3. Construction and properties of the proposed code set

In this section, we first present a new construction of OZCZ code set with parameters $(L, K, Z_{c z}) = (4 K Z_{c z}, K, Z_{c z})$ and $w = 2 K Z_{c z}$ . Then we derive and discuss the properties. Finally, an illustrative example of the generated code set with parameters $(L, K, Z_{c z}) = (32, 4, 2)$ and $w = 16$ is described.

3.1 Construction

For an ideal synchronous VLC system, the MAI among different users can be easily eliminated by adopting spreading codes with good correlation properties. However, due to the non-perfect clock, short time delay always exists in multi-users’ system. The OZCZ code set can be utilized in the QS-CDMA-VLC systems to solve this problem [16,17]. In this section, a new OZCZ code set can be constructed based on interleaving and iterating methods, as the following steps.

Step 1: A pair of initial codes $h_{1}^{(0)}$ and $h_{2}^{(0)}$ with length L = 8 written as

h_{1}^{(0)} = [h_{1, 0}, h_{1, 1}, ..., h_{1, L - 1}] = [- - - + + + - +],

h_{2}^{(0)} = [h_{2, 0}, h_{2, 1}, ..., h_{2, L - 1}] = [- + + - + + - -],

where ‘

+

’, ‘

-

’ denote ‘

+ 1

’ and ‘

- 1

’ respectively, and the periodic correlation function between

h_{1}^{(0)}

and

h_{2}^{(0)}

satisfy:

θ_{h_{i}^{(0)}, h_{j}^{(0)}} (τ) = {\begin{array}{l} 8 i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 1 \end{array} .

Step 2: Let n denote interleaving times, we can get

h_{1}^{(n)}

and

h_{2}^{(n)}

as follows,

h_{1}^{(n)} = I (h_{1}^{(n - 1)}, h_{2}^{(n - 1)}),

h_{2}^{(n)} = I (h_{1}^{(n - 1)}, - h_{2}^{(n - 1)}) .

Based on

h_{1}^{(n)}

and

h_{2}^{(n)}

, an initial matrix

H_{0}^{(n)} (2^{n + 3}, 2, 2^{n})

is obtained as

H_{0}^{(n)} = [\begin{matrix} h_{1}^{(n)} \\ h_{2}^{(n)} \end{matrix}] .

Step 3: The matrix

H_{0}^{(n)}

can be equally separated into four parts

[A_{0}^{(n)}]

,

[B_{0}^{(n)}]

,

[C_{0}^{(n)}]

and

[D_{0}^{(n)}]

as follows

H_{0}^{(n)} = [\begin{matrix} \begin{matrix} A_{0}^{(n)} & B_{0}^{(n)} \end{matrix} \\ \begin{matrix} C_{0}^{(n)} & D_{0}^{(n)} \end{matrix} \end{matrix}] .

Based on iterating method, a code set

H_{m}^{(n)} (2^{m + n + 3}, 2^{m + 1}, 2^{n})

can be constructed as follows

H_{m}^{(n)} = [\begin{matrix} \begin{matrix} A_{m}^{(n)} & B_{m}^{(n)} \end{matrix} \\ \begin{matrix} C_{m}^{(n)} & D_{m}^{(n)} \end{matrix} \end{matrix}] = [\begin{matrix} \begin{matrix} E \times A_{m - 1}^{(n)} & E \times B_{m - 1}^{(n)} \end{matrix} \\ \begin{matrix} F \times C_{m - 1}^{(n)} & F \times D_{m - 1}^{(n)} \end{matrix} \end{matrix}] m \geq 1,

where

E = [\begin{matrix} \begin{matrix} + \\ + \end{matrix} & \begin{matrix} - \\ + \end{matrix} \end{matrix}]

and

F = [\begin{matrix} \begin{matrix} + \\ + \end{matrix} & \begin{matrix} + \\ - \end{matrix} \end{matrix}]

are two 2-order Hadamard matrices,

m

is the iterating times.

Step 4: Based on the code set $H_{m}^{(n)}$ $(2^{m + n + 3}, 2^{m + 1}, 2^{n})$ , a pair of transmitting matrix T and receiving matrix R can be generated as follows,

\begin{matrix} O Z C Z = < R, T > \\ {\begin{cases} R = H_{m}^{(n)} = {_{r_{k}}^{}} \\ T = f (R) = {_{t_{k}^{d_{k}}}^{}} \end{cases}, \end{matrix}

where

f (\cdot)

denotes the mapping relation between

T

and

R

given as

t_{k}^{d_{k}} = \frac{1 + {(- 1)}^{d_{k}} r_{k}}{2},

where

d_{k} \in {0, 1}

is the

k -th

user input data.

Based on above steps, an OZCZ code set pair with parameters $(L, K, Z_{c z}) = (2^{m + n + 3}, 2^{m + 1}, 2^{n}) = (4 K Z_{c z}, K, Z_{c z})$ and $w = 2^{m + n + 2} = 2 K Z_{c z}$ can be generated.

3.2 The properties of the new construction

Theorem 1: The generated OZCZ code set has following properties:

1) The weight of each code is L/2.
2) The correlation properties can be presented as $θ_{r_{i}, t_{j}^{d_{k}}} (τ) = {\begin{matrix} {(- 1)}^{d_{k}} 2^{m + n + 2} i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2^{n} \end{matrix} n \geq 0, m \geq 0.$

Proof:

1) The initial codes $h_{1}^{(0)}$ and $h_{2}^{(0)}$ are both balanced codes. The codes after interleaving are still balanced. When applying iterating, the Hadamard matrices E and F are balanced too. It is easy to be proved that the weight of each generated code is L/2.
2) The periodic correlation function between $h_{1}^{(0)}$ and $h_{2}^{(0)}$ satisfy: $θ_{h_{i}^{(0)}, h_{j}^{(0)}} (τ) = {\begin{array}{l} 8 i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 1 \end{array},$

According to Definition 2, we have

θ_{h_{1}^{(n)}, h_{2}^{(n)}} (2 τ) = θ_{h_{1}^{(n - 1)}, h_{1}^{(n - 1)}} (τ) + θ_{h_{2}^{(n - 1)}, - h_{2}^{(n - 1)}} (τ),

θ_{h_{1}^{(n)}, h_{2}^{(n)}} (2 τ + 1) = θ_{h_{1}^{(n - 1)}, - h_{2}^{(n - 1)}} (τ) + θ_{h_{1}^{(n - 1)}, h_{2}^{(n - 1)}} (τ + 1) .

Consequently,

θ_{h_{i}^{(n)}, h_{j}^{(n)}} (τ) = {\begin{array}{l} 8 \times 2^{n} i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2^{n} \end{array} n \geq 0.

Because E and F are both Hadamard matrices, it is easily found that the iterating method only changes the code length L and the number of codes K while

Z_{c z}

keeping unchanged. Therefore, the code set

H_{m}^{(n)}

has ZCZ properties. The correlation properties of the receiving matrix

R = H_{m}^{(n)} = {r_{k}}_{k = 1}^{2^{m + 1}}

can be derived as

θ_{r_{i}, r_{j}} (τ) = {\begin{array}{l} 2^{m + n + 3} i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2^{n} \end{array} n \geq 0, m \geq 0.

In the new construction, the weight of each code is L/2 and each transmitting code satisfies

t_{k}^{d_{k}} = \frac{1 + {(- 1)}^{d_{k}} r_{k}}{2}

. The correlation function for the new construction can be simplified as follows

θ_{r_{i}, t_{j}^{d_{k}}} (τ) = \sum_{l = 0}^{L - 1} r_{i, l} t_{j, (l + τ) \mod L}^{d_{k}} = \sum_{l = 0}^{L - 1} r_{i, l} (\frac{1 + {(- 1)}^{d_{k}} r_{j, (l + τ) \mod L}}{2}) = {(- 1)}^{d_{k}} \frac{θ_{r_{i}, r_{j}} (τ)}{2} .

Consequently, the PACF and PCCF correlation function for the new construction can be given by

θ_{r_{i}, t_{j}^{d_{k}}} (τ) = {\begin{matrix} {(- 1)}^{d_{k}} 2^{m + n + 2} i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2^{n} \end{matrix} n \geq 0, m \geq 0.

Based on above analysis, the new construction of OZCZ code set including a pair of unipolar and bipolar code sets satisfies the ideal auto-correlation and cross-correlation properties within

Z_{c z} = 2^{n}

, which is helpful to reduce the effect of time delay between different users. The length of zero correlation zone can easily extend by interleaving, compared with the construction in [17]. In a VLC-CDMA system using the proposed OZCZ code, the transmitter employs the unipolar code set for the intensity modulation directly and the receiver employs the bipolar code set for the received signals after optical-electrical conversion. Furthermore, compared with the construction in [16,19], the weight of the new generated codes maintains 50% that indicates the dimming values of 50% can be obtained regardless of user numbers, which can improve significantly the illumination performance of the CDMA-VLC system.

3.3 Example of the new construction

There is an example of the proposed OZCZ code set having parameters $(L, K, Z_{c z}) = (32, 4, 2)$ and $w = 16$ by once interleaving and iterating. The generation steps and properties are given as follows.

Step 1: For $n = 0$ ,

H_{0}^{(0)} = [\begin{matrix} h_{1}^{(0)} \\ h_{2}^{(0)} \end{matrix}] = [\begin{matrix} - - - + + + - + \\ - + + - + + - - \end{matrix}] .

For

n = 1

,

H_{0}^{(1)} = [\begin{matrix} h_{1}^{(1)} \\ h_{2}^{(1)} \end{matrix}] = [\begin{matrix} I (h_{1}^{(0)}, h_{2}^{(0)}) \\ I (h_{1}^{(0)}, - h_{2}^{(0)}) \end{matrix}] = [\begin{matrix} - - - + - + + - + + + + - - + - \\ - + - - - - + + + - + - - + + + \end{matrix}] .

The correlation properties satisfy

θ_{h_{i}^{(1)}, h_{j}^{(1)}} (τ) = {\begin{array}{l} 16 i = j, τ = 0 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2 \end{array} .

Step 2: For

m = 0

, the matrix

H_{0}^{(1)}

can be separated into four parts as follows

H_{0}^{(1)} = [\begin{matrix} \begin{matrix} A_{0}^{(1)} & B_{0}^{(1)} \end{matrix} \\ \begin{matrix} C_{0}^{(1)} & D_{0}^{(1)} \end{matrix} \end{matrix}] = [\begin{matrix} \begin{matrix} [- - - + - + + -] & [+ + + + - - + -] \end{matrix} \\ \begin{matrix} [- + - - - - + +] & [+ - + - - + + +] \end{matrix} \end{matrix}] .

By once iterating, the code set

H_{1}^{(1)} (32, 4, 2)

can be generated by

\begin{matrix} H_{1}^{(1)} = [\begin{matrix} \begin{matrix} A_{1}^{(1)} & B_{1}^{(1)} \end{matrix} \\ \begin{matrix} C_{1}^{(1)} & D_{1}^{(1)} \end{matrix} \end{matrix}] = [\begin{matrix} \begin{matrix} E \times A_{0}^{(1)} & E \times B_{0}^{(1)} \end{matrix} \\ \begin{matrix} F \times C_{0}^{(1)} & F \times D_{0}^{(1)} \end{matrix} \end{matrix}] \\ = [\begin{matrix} - - - + - + + - + + + - + - - + + + + + - - + - - - - - + + - + \\ - - - + - + + - - - - + - + + - + + + + - - + - + + + + - - + - \\ - + - - - - + + - + - - - - + + + - + - - + + + + - + - - + + + \\ - + - - - - + + + - + + + + - - + - + - - + + + - + - + + - - - \end{matrix}] . \end{matrix}

Step 3: The new construction OZCZ for a 4-user system can be obtained by a combination of the receiving matrix R and the transmitting matrix T.

\begin{matrix} O Z C Z = < R, T > \\ {\begin{cases} R = H_{1}^{(1)} = {_{r_{k}}^{}} \\ T = f (R) = {_{t_{k}^{d_{k}}}^{}} = {\begin{matrix} [\begin{matrix} 000 + 0 + + 0 + + + 0 + 00 + + + + + 00 + 00000 + + 0 + \\ 000 + 0 + + 0000 + 0 + + 0 + + + + 00 + 0 + + + + 00 + 0 \\ 0 + 0000 + + 0 + 0000 + + + 0 + 00 + + + + 0 + 00 + + + \\ 0 + 0000 + + + 0 + + + + 00 + 0 + 00 + + + 0 + 0 + + 000 \end{matrix}] d_{k} = 0 \\ [\begin{matrix} + + + 0 + 00 + 000 + 0 + + 00000 + + 0 + + + + + 00 + 0 \\ + + + 0 + 00 + + + + 0 + 00 + 0000 + + 0 + 0000 + + 0 + \\ + 0 + + + + 00 + 0 + + + + 000 + 0 + + 0000 + 0 + + 000 \\ + 0 + + + 000 + 0000 + + 0 + 0 + + 000 + 0 + 00 + + + \end{matrix}] d_{k} = 1 \end{matrix} \end{cases} . \end{matrix}

The correlation functions for the codes in the matrix T and R satisfy:

θ_{r_{i}, t_{j}^{d_{k}}} (τ) = {\begin{array}{l} 16 i = j, τ = 0, d_{k} = 0 \\ - 16 i = j, τ = 0, d_{k} = 1 \\ 0 i \neq j, τ = 0 \\ 0 0 < | τ | \leq 2 \end{array} n = 1, m = 1.

4. The QS-CDMA-VLC system design and experiment setup

Figure 1 gives the QS-CDMA-VLC system design using the new OZCZ construction, which is used in the subsequent performance analysis. For simplicity, we only focus on the point-to-point transmission in LOS (Line-of-sight) case. In the transmitter, the k-th user original data $d_{k} (t)$ is first through spread module. Different from most of traditional CDMA-VLC system, the spread module is an accurate mapping operation. That is, when the $k -th$ user’s data $d_{k} = 0$ , the data after spreading is $t_{k}^{0}$ , otherwise $t_{k}^{1}$ . The total transmitted optical signal $s (t)$ can be written as

s (t) = \sum_{k = 1}^{K} s_{k} (t) = \sum_{k = 1}^{K} t_{k}^{d_{k} (t)} (t) 0 \leq t \leq T = L T_{c},

where

s_{k} (t)

is the

k -th

user transmitted signal,

T_{c}

is the chip time interval and

T = L T_{c}

is the symbol period. The proposed OZCZ code waveform

t_{k}^{d_{k} (t)} (t)

for

k -th

user transmitting one symbol

d_{k} (t)

can be given by [16]

t_{k}^{d_{k} (t)} (t) = \sum_{i = 0}^{L - 1} \frac{1 + {(- 1)}^{d_{k} (t)} r_{k} (i)}{2} P_{T_{c}} (t - i T_{c}),

where

r_{k} (i) = r_{k} (i + L) \in {- 1, 1}

, and

P_{T_{c}} (t)

is a unit rectangular pulse of duration

T_{c}

. At the receiver, the received signal

r (t) = s (t) + n (t)

from the photodetector (PD) contains the active users’ signals and the additive white Gaussian noise (AWGN) signals. For the k-th user, the decision threshold

T h

can be obtained as follows

\begin{matrix} T h = r_{k} (t) \times r (t) \\ = \sum_{i = 0}^{L - 1} r_{k} (i) P_{T_{c}} (t - i T_{c}) \times (\sum_{k = 1}^{K} \sum_{i = 0}^{L - 1} \frac{1 + {(- 1)}^{d_{k} (t)} r_{k} (i)}{2} P_{T_{c}} (t - i T_{c}) + n (t)) . \end{matrix}

After de-spreading, the power of the noise signal n(t) is small enough to be ignored [6] and the direct component can be easily filtered. Based on the above construction steps, the proposed codes of each user satisfy

\sum_{i = 0}^{L - 1} r_{k_{1}} (i) r_{k_{2}} (i) = {\begin{matrix} 4 K Z_{c z} k_{1} = k_{2} \\ 0 k_{1} \neq k_{2} \end{matrix} .

And the correlation properties of each code would not be changed when the time delay does not exceed zone correlation zone length of the construction. Finally, the desired user data

d_{k} (t)

can be obtained by

Fig. 1 Block diagram of the QS-CDMA-VLC system.

Download Full Size | PDF

d_{k} (t) = {\begin{matrix} 0 T h \geq 0 \\ 1 T h < 0 \end{matrix} .

Figure 2 shows the experimental setup used for the QS-CDMA experimental transmission system. In the experiments, we investigate the BER performance of the K-user CDMA-VLC system based on the proposed OZCZ code for different sampling rate of transmitted signals, number of delay users and free-space transmission distance. The transmitted signals s(t) with different users were generated by an arbitrary waveform generator (AWG 70002A) and then amplified by an amplifier (ZHL-6A-S + ). A single red LED as the light source was driven by a bias-Tee (ZFBT-6GW + ) and optimally biased at 3.5V. At the receiver, the optical signal r(t) was detected by a PD (APD AD500) and then recorded by a mixed signal oscilloscope (MSO 70604C) for offline signal processing. The sampling rate of MSO was 625 MS/s, while the sampling rate of AWG varied from 150 MS/s to 250 MS/s. All the experiments were conducted under normal ambient light (~200lux).

Fig. 2 Experiment setup of the QS-CDMA-VLC system.

Download Full Size | PDF

5. Results and discussion

The simulation and experimental results of the proposed construction for the QS- CDMA-VLC system model are provided in the following. We evaluate the system performance with different user numbers, sample rates of AWG, delay user numbers and transmission distances. The length of spreading codes would affect the transmission efficiency of CDMA system. With the fixed number of users and the minimum length of zero correlation zone, each code can achieve the minimum length. We adopt the new construction with $Z_{c z} = 1$ to ensure the transmission efficiency in simulations and experiments. For simplicity, we assume that the time delay does not exceed the zero correlation zone length. In all the result figures, the FEC threshold of 3.8 × 10⁻³ is shown by a horizontal line.

Figure 3 presents the BER performance of the CDMA-VLC system using the proposed OZCZ codes with different active user numbers. In the experiment, the free-space transmission distance is set to 1 m and the sample rate of AWG is adjusted from 150 MS/s to 250 MS/s. The simulation and experimental results show that the BER performance of the system is degraded as the user numbers increase for all the sample rates. The reason is that with the increase of active user numbers, a single user cannot effectively suppress MAI, which will limit the capacity of the system. It is demonstrated that due to the influence of high-frequency fading characteristics of LEDs, the transmitted signals have a lower power with the higher sample rate, leading to the decrease of system performance as shown in Fig. 3. The further improvement of transmission performance can be realized by channel equalization technology.

Fig. 3 (a) Simulation results and (b) experiment results of BER performance versus the number of active users for the CDMA-VLC system.

Download Full Size | PDF

We further investigate the impact of the delay user number on the BER performance of the 32-user QS-CDMA-VLC system. Because the worst condition in the system is that half of users have a time delay $τ = 1$ , the number of delay users in the simulation and experiment is set as 0, 4, 8, 12 and 16. As we can see from Fig. 4, the BER performance of the system using the proposed code construction varies in an acceptable range even for 16 delay users from both simulation and experiment results. The reason is that when the time delay is within zero correlation zone, each user still has a unique spreading code with good correlation properties to obtain the desired signals. It is also demonstrated that from Fig. 4 the increase of sample rate leads to the BER performance degradation, which is similar with above discussion about system performance with different user numbers. As a result, the sample rate of 250 MS/s over 1 m transmission is successfully achieved with the BER near FEC for a QS-CDMA-VLC system. That is, it can effectively avoid small time delay among users induced by the non-perfect clock. Furthermore, if the synchronization condition among users gets worse, we can extend the length of zero correlation zone to tolerate larger time delay.

Fig. 4 (a) Simulation results and (b) experiment results of BER performance of the 32-user QS-CDMA VLC system versus a number of delay users.

Download Full Size | PDF

The BER performance of the experimental CDMA-VLC system with different active users versus free-space transmission distance is further evaluated as shown in Fig. 5. The sample rate of AWG is set to 250 MS/s and the number of users varies from 4 to 32 users. The experimental results reveal that system MAI can be suppressed effectively due to the good correlation properties of the proposed code structure as shown in Fig. 5. However, there is still BER degradation with the increasing number of users as mentioned above. Furthermore, with the increase of free-space transmission distance, the BER performances with different number of active users have a similar degradation trend. This is because when the input voltage of LED is fixed, the input power of PD decreases as the increase of transmission distance. It is shown that the 16-user system achieved a BER of 3.7 × 10⁻³ for a 2.0 m transmission while the 32-user system achieved a BER of 2.7 × 10⁻³ for a 1.5 m transmission in order to satisfy the FEC limit.

Fig. 5 BER performance of the experimental system versus free-space transmission distance with different active users.

Download Full Size | PDF

6. Conclusion

In this paper, a new construction of OZCZ code set with parameters $(L, K, Z_{c z}) = (4 K Z_{c z}, K, Z_{c z})$ and $w = 2 K Z_{c z}$ is proposed based on interleaving and iterating methods. The proposed code sets have ideal zero correlation zone properties and can be more suitable for the QS VLC system. The illumination performance can also be enhanced. We further investigate and analyze the BER performance of the QS-CDMA-VLC system employing the proposed OZCZ codes theoretically and experimentally. The results show that MAI can be suppressed effectively for the CDMA-VLC system with different user numbers, sample rates, delay user numbers and transmission distances. The simulation and experimental results demonstrate for the 32-user system, about 250 MS/s sample rate of can successfully achieve for the FEC limit at a distance of 1.5 m. Furthermore, when the time delays among users do not exceed the zero correlation zone length, the BER varies slightly in an acceptable range. It is revealed that the new proposed OZCZ code set can be considered as a potential and suitable candidate for multiple access in VLC system, which can significantly overcome non-perfect synchronous problem.

Funding

National Natural Science Foundation of China (61671055).

References

1. S. Wu, H. Wang, and C. H. Youn, “Visible light communications for 5G wireless networking systems: from fixed to mobile communications,” IEEE Netw. 28(6), 41–45 (2014). [CrossRef]

2. H. Burchardt, N. Serafimovski, D. Tsonev, S. Videv, and H. Haas, “VLC: Beyond point-to-point communication,” IEEE Commun. Mag. 52(7), 98–105 (2014). [CrossRef]

3. Y. Qiu, S. Chen, H. H. Chen, and W. Meng, “Visible light communications based on CDMA technology,” IEEE Wirel. Commun. 25(2), 178–185 (2018). [CrossRef]

4. Z. Zheng, T. Chen, L. Liu, and W. Hu, “Experimental demonstration of femtocell visible light communication system employing code division multiple access,” in 2015 Optical Fiber Communications Conference and Exhibition (OFC) (Optical Society of America, 2015), pp. 1–3.

5. S. H. Chen and C. W. Chow, “Color-shift keying and code-division multiple-access transmission for RGB-LED visible light communications using mobile phone camera,” IEEE Photonics J. 6(6), 1–6 (2014). [CrossRef]

6. J. An and W.-Y. Chung, “Single cell multiple-channel VLC with 3-level amplitude-based CDMA,” Opt. Commun. 432, 13–19 (2019). [CrossRef]

7. M. Hammouda, A. M. Vegni, J. Peissig, and M. Biagi, “Resource allocation in a multi-color DS-OCDMA VLC cellular architecture,” Opt. Express 26(5), 5940–5961 (2018). [CrossRef] [PubMed]

8. S. Pergoloni, M. Biagi, S. Colonnese, R. Cusani, and G. Scarano, “Optimized LEDs footprinting for indoor visible light communication networks,” IEEE Photonics Technol. Lett. 28(4), 532–535 (2016). [CrossRef]

9. H. Qian, S. Dai, S. Zhao, S. Cai, and H. Zhang, “A robust CDMA VLC system against front-end nonlinearity,” IEEE Photonics J. 7(5), 1–9 (2015). [CrossRef]

10. Y. A. Chen, Y. T. Chang, Y. C. Tseng, and W. T. Chen, “A framework for simultaneous message broadcasting using CDMA-based visible light communications,” IEEE Sens. J. 15(12), 6819–6827 (2015). [CrossRef]

11. D. Chen, J. Wang, J. Jin, H. Lu, and L. Feng, “A CDMA system implementation with dimming control for visible light communication,” Opt. Commun. 412, 172–177 (2018). [CrossRef]

12. C. He, L. Yang, P. Xiao, and M. A. Imran, “DS-CDMA assisted visible light communications systems,” in 20th International Workshop on Computer Aided Modelling and Design of Communication Links and Networks (CAMAD), (IEEE,2015), pp. 27–32.

13. M. Noshad and M. Brandtpearce, “Application of Expurgated PPM to indoor visible light communications—Part I: single-user systems,” J. Lightwave Technol. 32(5), 875–882 (2014). [CrossRef]

14. M. F. Guerra-Medina, O. Gonzalez, B. Rojas-Guillama, J. A. Martin-Gonzalez, F. Delgado, and J. Rabadán, “Ethernet-OCDMA system for multi-user visible light communications,” Electron. Lett. 48(4), 227–228 (2012). [CrossRef]

15. T. K. Matsushima, S. Sasaki, M. Kakuyama, S. Yamasaki, and Y. Murata, “A visible-light communication system using optical CDMA with inverted MPSC,” in 6th International Workshop on Signal Design Its Applications in Communications, (IEEE,2013), pp. 52–55. [CrossRef]

16. L. Feng, J. Wang, R. Q. Hu, and L. Li, “New design of optical zero correlation zone codes in quasi-synchronous VLC CDMA systems,” EURASIP J. Wirel. Commun. Netw. 2015(1), 120 (2015). [CrossRef]

17. B. Fassi and A. Taleb-Ahmed, “A new construction of optical zero-correlation zone codes,” J. Opt. Commun. 39(3), 359–368 (2018). [CrossRef]

18. G. Ahmed and A. Djebbari, “New technique for construction of a zero cross correlation code,” Optik (Stuttg.) 123(15), 1382–1384 (2012). [CrossRef]

19. M. S. Anuar, S. A. Aljunid, N. M. Saad, and S. M. Hamzah, “New design of spectral amplitude coding in OCDMA with zero cross-correlation,” Opt. Commun. 282(14), 2659–2664 (2009). [CrossRef]

20. K. Kim, K. Lee, and K. Lee, “Appropriate RLL coding scheme for effective dimming control in VLC,” Electron. Lett. 52(19), 1622–1624 (2016). [CrossRef]

21. L. Feng, R. Q. Hu, J. Wang, and X. Peng, “Fountain code-based error control scheme for dimmable visible light communication systems,” Opt. Commun. 347, 20–24 (2015). [CrossRef]

22. L. Wang and X. Zeng, “A new class of sequences with zero correlation zone based on interleaving perfect sequence,” in Proceedings of the 5th International Workshop on Signal Design and Its Applications in Communications, (IEEE,2011), pp. 29–31. [CrossRef]

Experimental demonstration of quasi-synchronous CDMA-VLC systems employing a new OZCZ code construction

Abstract

1. Introduction

2. Preliminaries

3. Construction and properties of the proposed code set

3.1 Construction

3.2 The properties of the new construction

3.3 Example of the new construction

4. The QS-CDMA-VLC system design and experiment setup

5. Results and discussion

6. Conclusion

Funding

References

Cited By

Figures (5)

Equations (36)

Optics Express