OTFS wireless data energy simultaneous transmission method based on frequency domain equalizer in high maneuvering environment

Abstract

The invention relates to the orthogonal time-frequency-space field of wireless communication, in particular to an OTFS wireless data-energy simultaneous transmission method based on a frequency domain equalizer in a high-mobility environment. A wireless information and energy simultaneous transmission technology is applied to an OTFS transmission system, a power distribution and frequency domain linear equalizer is introduced, a strict mathematical optimization model is established according to a system model and an SWIPT power division structure, an optimal power division factor and an optimal power distribution scheme are solved by using a convex optimization method, and the optimal power division factor and the optimal power distribution scheme are obtained. The objective of the invention is to maximize a partial signal-to-noise ratio or a signal-to-signal plus interference plus noise ratio of a receiving end of a system for information decoding. Simulation results show that the proposed optimal power allocation scheme obtains lower bit error rate performance, and meanwhile, compared with an OFDM transmission system, it is proved that OTFS transmission has better system performance in a high-speed mobile environment.

Description

Technical Field

[0001] The present invention relates to the orthogonal time frequency space (OTFS) field of wireless communication and the simultaneous wireless information and power transfer (SWIPT) technology. Background Art

[0002] The sixth generation of mobile communication technology (6G) has the characteristics of high speed, low latency and large capacity. The superior performance of 6G communication compared with the previous generation of mobile communication has driven the development, implementation and application of many emerging technologies, such as unmanned aerial vehicles (UAV), vehicle-to-everything communications (V2X) and high-speed railway (HSR). Among them, the wireless channel is usually time-varying and has severe Doppler spread. In this case, the reliability of the transmission link will face huge challenges. For example, in the orthogonal frequency division multiplexing (OFDM) scheme, high Doppler shift can easily destroy the orthogonality between subcarriers, resulting in severe inter-carrier interference (ICI), and the communication quality will also be reduced.

[0003] Delay-Doppler (DD) domain modulation is a communication method that is robust to both Doppler spreading and delay spreading dual-dispersion channels. Time-varying wireless multipath channels have a sparse representation in the DD domain. In order to cope with time-varying channels, some scholars have proposed a new modulation technology - Orthogonal Time Frequency Space (OTFS) modulation technology. Specifically, a two-dimensional orthogonal modulation process is used to convert the dual-dispersion channel in the time-frequency domain (TF) into an approximate time-invariant channel in the delay-Doppler domain. Compared with the TF modulation scheme, OTFS can simultaneously utilize time and frequency diversity, significantly reducing the complexity of time-varying channel estimation.

[0004] Meanwhile, in the upcoming 6G system, the Internet of Things (IoT) plays an important role, such as smart cities, autonomous driving, wearable devices, and healthcare. However, with the growing demand for IoT applications, replacing or charging batteries for IoT devices may be inconvenient or even impossible. To this end, various energy harvesting (EH) technologies have been used to extend the working time of IoT networks, for example, harvesting energy from solar energy, wind energy, and ambient RF signals. RF EH technology can be divided into two categories: wireless power transfer (WPT) and simultaneous wireless information and power transfer (SWIPT). Compared with pure WPT, SWIPT can bring more practical properties because it can provide energy and information access to wireless devices at the same time. However, the system design and optimization of SWIPT is usually more complicated than that of WPT systems because it requires balancing EH performance and information decoding.

[0005] (Information Decoding, ID) performance. Therefore, it has triggered a considerable research boom. Power splitting (PS) is a method to implement SWIPT, which divides the signal received by the receiving antenna into two parts, one for ID and the other for EH.

[0006] Tie et al. studied the optimal power allocation for OTFS transmission using frequency domain linear equalizers, including frequency domain zero-forcing (ZF) and minimum mean squared error (MMSE) equalizers. A closed-form expression for the received signal-to-noise ratio was derived, and the optimization problem of maximizing the received signal-to-noise ratio was solved by applying the Lagrange multiplier method. Peng et al. studied a joint transceiver design and receive power partitioning optimization for a MIMO-SWIPT system. The harvested power of the receiver was maximized under the MSE constraint and the total transmit power constraint. First, the linear MMSE equalization matrix was derived, and a non-convex optimization problem involving the transmit precoding matrix and the receive PS ratio was obtained. For a given PS ratio, it was proved that the problem can be transformed into a semidefinite programming problem, which can be solved using an interior point algorithm. Then, using the unimodality of the objective function for the PS ratio, a golden section search method was proposed to effectively find the optimal PS ratio. Furthermore, in order to reduce the complexity of solving the internal SDP problem, the optimal structure of the precoding matrix was used to transform the original matrix-based optimization problem into a scalar-based optimization problem, and a closed-form solution was obtained through convex optimization techniques. Summary of the invention

[0007] The purpose of this invention is to fill the gap in current research. The research object is mainly the OTFS-SWIPT system, so as to make full use of the robustness brought by OTFS, the convenience of simultaneous transmission of data and energy brought by SWIPT, and the gain of power allocation for signal counteracting the influence of channel interference.

[0008] The technical solution adopted by the present invention is:

[0009] An OTFS wireless data and energy simultaneous transmission method based on a frequency domain equalizer in a high mobility environment includes the following processes:

[0010] Step 1: At the transmitting end, the input DD domain signal is transformed into the TF domain by inverse symplectic finite Fourier transform;

[0011] Step 2, applying the power allocation matrix to the TF domain signal to obtain a signal after power allocation;

[0012] Step 3, perform Heisenberg transform on the power-allocated signal to obtain a time domain signal, which is sent out through the channel;

[0013] Step 4, at the receiving end, the received time domain signal is then power split;

[0014] Step 5: The power-segmented signal is converted back to the TF domain through Wigner transformation, and energy is collected at the same time;

[0015] Step 6: Perform frequency domain equalization on the TF domain signal, and then perform symplectic finite Fourier transform, that is, convert it back to DD domain signal.

[0016] Furthermore, the frequency domain equalization is processed using a frequency domain linear equalizer, including a ZF equalizer and an MMSE equalizer.

[0017] Among them, when the ZF equalizer is used for frequency domain equalization processing, the following optimization problem is established for the power allocation and power division parameters:

[0018]

[0019] in,

[0020] In the formula, p nM+m is the diagonal element in the power allocation matrix P, M and N are the number of subcarriers and time slots respectively, P is a diagonal matrix of size NM*NM, λ is the division factor of power division, H[n,m] is the channel response in the TF domain, ζ is the energy conversion efficiency, e is the minimum power required for energy collection EH, ρ is the variance of the symbol, P T is the maximum transmission power allowed by the system, is the noise power received by the receiving antenna, σ 2 The additional noise power introduced when the receiving end decodes the information;

[0021] Solving the optimization problem obtains the power allocation matrix P and the power splitting factor λ.

[0022] Among them, when the MMSE equalizer is used for frequency domain equalization processing, the following optimization problem is established for the power allocation and power splitting parameters:

[0023]

[0024] in,

[0025] In the formula, p nM+m is the diagonal element in the power allocation matrix P, M and N are the number of subcarriers and time slots respectively, P is a diagonal matrix of size NM*NM, λ is the division factor of power division, H[n,m] is the channel response in the TF domain, ζ is the energy conversion efficiency, e is the minimum power required for energy collection EH, ρ is the variance of each symbol, P T is the maximum transmission power allowed by the system, is the noise power received by the receiving antenna, σ 2 The additional noise power introduced when the receiving end decodes the information;

[0026] Solving the optimization problem obtains the power allocation matrix P and the power splitting factor λ.

[0027] The advantages of the present invention compared with the prior art are:

[0028] The object of the present invention is mainly the OTFS-SWIPT system, and the purpose is to fill the gap in current research, because the number of studies on this system at home and abroad is very small. OTFS is also considered to be one of the options for future communications due to its robustness to high Doppler frequency shift in high-speed scenarios. SWIPT can transmit signals and energy at the same time, which can reduce the cost of wires and wiring, eliminate the trouble of replacing batteries for wireless devices, and extend the service life of the equipment. In addition, since OTFS has a more complex modulation process than OFDM, with conversion calculations from DD domain to TF domain, etc., there is little research on OTFS power allocation. Therefore, the present invention currently focuses on the research on the power allocation problem of OTFS-SWIPT. The research on power allocation can enable the signal to more effectively resist the interference caused by the channel, thereby improving the performance of the communication system, especially the reliability.

[0029] Therefore, the significance of the research of the present invention is mainly to make full use of the robustness brought by OTFS, the convenience of simultaneous transmission of data and energy brought by SWIPT, and the gain of power allocation for signal counteracting the influence of channel interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the OTFS-SWIPT system flow of the present invention.

[0031] Figure 2 A comparison chart of BER-SNR curves of the present invention and OFDM system using OPA and EPA.

[0032] Figure 3 This is a comparison diagram of BER-SNR curves of noise introduced by different IDs when OPA is applied in the present invention.

[0033] Figure 4 This is a comparison diagram of BER-SNR curves for different channel multipath numbers when OPA is applied in the present invention. DETAILED DESCRIPTION

[0034] The present invention applies SWIPT technology to the OTFS system. First, a rough model of the OTFS-SWIPT system is designed, and the relationship between the sending and receiving signals is derived in terms of formulas to obtain its expression. Then, a frequency-domain linear equalizer (Frequency-Domain Linear Equalizers) is introduced in the time-frequency domain to equalize the received signal to compensate for channel distortion and reduce inter-symbol interference (ISI), thereby improving the quality of the received signal. It includes a zero-forcing equalizer and a minimum mean square error equalizer. Then, the matrix of signal power allocation is derived and its optimization problem is obtained. The signal-to-noise ratio (SNR) of the signal recovered from the partial signal used for information decoding at the receiving end is obtained, and the optimization goal is to maximize the SNR. The constraints are the total power constraint of the transmission and the minimum energy requirement required to be provided by the part used to collect energy in the SWIPT system, as well as the range of the power allocation factor, the non-negative constraint of energy, etc.

[0035] 1. System model

[0036] Consider a system model of a point-to-point transmission OTFS scheme in which the transmitter moves at high speed while the receiver remains stationary, and the transmission is performed with a single antenna. Assume that within an OTFS frame, the time duration is NT (s) and the bandwidth is MΔf (Hz), where the subcarrier spacing Δf = 1 / T, M and N are the number of subcarriers and the number of time slots, respectively. Assume that an OTFS frame consists of MN symbols {x[k,l], k = 0, ..., N-1, l = 0, ..., M-1} arranged in the DD domain. It is transformed into the TF domain by applying ISFFT (Inverse Symplectic FiniteFourier Transform):

[0037]

[0038] Among them, the symbol represents the Kronecker product operator, x d =[x[0,0],…x[0,M-1],…,x[N-1,0],…x[N-1,M-1]] T , similarly, the time-frequency signal x f =[X[0,0],…X[0,M-1],…,X[N-1,0],…X[N-1,M-1]] T , X[n,m] is the signal in the TF domain. N represents the N-point normalized DFT matrix, F M represents the normalized DFT matrix of point M. Assume that the variance of each symbol is E{x[k,l]| 2}=ρ. Apply the power allocation matrix to the TF domain signal, that is, Px f , where P is a diagonal matrix of size NM*NM, where each diagonal element is p nM+m After applying the Heisenberg transform to the signal after power allocation, the time domain signal s(t) can be obtained and sent out.

[0039] Assuming that the OTFS transmission signal goes through a dual-selective channel with Q independent paths, the channel response can be expressed as:

[0040]

[0041] Where T is the duration of each OFDM symbol (OTFS can be regarded as a continuous OFDM symbol), δ is the Dirac function (recognized), is the Doppler shift, h q , τ q and denote the channel coefficient, delay and Doppler shift on the qth path respectively. Assume that h qIt obeys Rayleigh distribution. The channel response in TF domain can be expressed as:

[0042]

[0043] The time domain signal r(t) received by the receiver can be converted back to the TF domain through Wigner transform. In order to facilitate the solution of power allocation, it is assumed that the receiving and transmitting pulses satisfy the biorthogonal property. Then the TF domain received signal can be expressed as:

[0044] y f =HPx f +z f (4)

[0045] Where H is a diagonal matrix of size NM*NM, and the diagonal elements are H[n,m]. f is a Gaussian white noise in the TF domain, whose elements are Z[n,m], assuming that it satisfies Due to the introduction of the SWIPT system, the power of the received signal is split. Assuming the split factor is λ, the part used for information decoding ID is The part used for energy harvesting EH is Then use the frequency domain equalizer to get y on the TF domain signal of ID, and then perform SFFT on it to convert it back to the DD domain to get y d :

[0046]

[0047] 2. Optimization Problem Model

[0048] According to the difference of frequency domain linear equalizer, it can be divided into ZF and MMSE. First, let’s look at the case of using ZF equalizer. -1 H -1 Applied to the TF domain signal, we get:

[0049]

[0050] Where n f It represents the additional noise brought by ID, which has a mean of 0 and a covariance matrix of σ 2 I is a cyclically symmetric complex Gaussian distribution. Then the SFFT is performed on it, that is, multiplying it by

[0051]

[0052] Among them, the noise terms are introduced from the channel and the ID. Therefore, they are independent of each other and can be separated during calculation. For example, when calculating the noise power introduced by the ID, the covariance matrix is:

[0053]

[0054] Because P -1 H -1 H -H P -H is a diagonal matrix, so the diagonal elements of the covariance matrix are and After the two matrices are multiplied, the diagonal elements are equal, which can be recorded as

[0055]

[0056] The equivalent SNR after equalization can be expressed as:

[0057]

[0058] Where ρ is the variance of each symbol, is the noise power received by the receiving antenna, σ 2 It is the additional noise power introduced when the receiving end decodes the information.

[0059] Similarly, the power that can be obtained from the signal used for energy harvesting EH is:

[0060]

[0061] Since the power of noise is much smaller than the signal power, it is ignored for the sake of simplicity. ζ represents the energy conversion efficiency.

[0062] The optimization problem is as follows:

[0063]

[0064] Among them, the second constraint is the total allocated transmission power constraint, the third constraint is the most basic EH collection power constraint required by the system, and e is the minimum power required by EH. It can be seen that through this special OTFS model, the optimization problem can be transformed from matrix operation to numerical solution of multiple scalars. There are many ways to solve this problem, and two are given here. The first one divides the problem into an inner and outer layer, with a fixed power division factor λ for the inner layer, and solves about It can be seen from the optimization problem that the inner layer is obviously a convex problem, so the numerical solution can be directly solved using MATLAB's CVX toolkit. The outer layer is a unimodal function of the power division factor λ, and its unimodal nature can be seen from the optimization problem. In order to maximize the SNR, the power division factor λ must be as large as possible. However, the energy constraint of the constraint requires that λ should be less than a value, otherwise the required energy will not be collected. Therefore, to solve this outer layer problem, a one-dimensional search method can be used directly, such as the Golden section search. Method 2 is also divided into inner and outer layer problems. The difference is that to solve the inner layer problem, the Lagrange multiplier method can be used directly to obtain the Lagrangian function:

[0065]

[0066] where μ1 and μ2 are Lagrange multipliers, and Take the derivative and set its derivative function to 0, and then sort it out to get:

[0067]

[0068] The values ​​of the two Lagrange multipliers can be solved by the subgradient method, so the power can be solved in the end.

[0069] Next, in order to compare with the ZF equalizer, we will analyze the situation of using the MMSE equalizer in this system, that is, the equalization matrix Applied to the TF domain signal, we get:

[0070]

[0071] Then perform SFFT conversion back to DD domain, that is, multiply by

[0072]

[0073] Then the equivalent signal-to-signal-plus-interference-plus-noise ratio (SSINR) after equalization can be calculated in the same way as ZF, expressed as:

[0074]

[0075] The signal used for energy harvesting EH is still the same as (8) because it is not equalized. In summary, the optimization problem is as follows:

[0076]

[0077] It can be found that only the objective function is different from the optimization problem 1, and the other constraints remain unchanged. Obviously, this problem can still be solved by dividing it into inner and outer problems. The inner problem is still a convex problem, so its numerical solution can also be easily obtained through MATLAB library functions, such as CVX. Of course, the Lagrange multiplier method can also be used to solve it, just like the ZF case.

[0078] 3. Simulation results

[0079] The present invention compares the change curve of system BER-SNR under different power allocation schemes, different ID-introduced noise sizes and channel multipath numbers. The basic experimental parameters of the simulation are shown in Table 1, where Jakes' formula is used to generate Doppler frequency shift according to the given speed, the path delay is uniformly distributed on [0, T], the channel path gain obeys a cyclic symmetric complex Gaussian distribution with a mean of zero and a variance of 1 / Q, and 4QAM modulation is applied to each symbol. It is assumed that the channel matrix can be perfectly estimated.

[0080] Table 1 Simulation parameters of SISO-OTFS-SWIPT system

[0081]

[0082]

[0083] 3.1 Comparison of BER-SNR curves when different power allocation schemes are applied to this system and OFDM system:

[0084] At this time, the additional noise power introduced by the ID is set to be equal to the noise power brought by the channel. The number of channel multipaths Q = 5. The simulation results show that compared with the equal power allocation (EPA), that is, the power allocated to each symbol is equal, the optimal power allocation (OPA) of the proposed optimization problem can improve the bit error rate performance of the communication system, that is, under the same SNR, the system using the OPA scheme has a lower BER than the system using the EPA scheme, especially in the high signal-to-noise ratio area, such as Figure 2As shown. It can also be seen that compared with the FD-MMSE equalizer, the FD-ZF equalizer is more sensitive to the power allocation in the TF domain because it reduces the impact of bad sub-channels, of course, at the cost of complexity. At the same time, it can also be seen that the OTFS system generally has a lower bit error rate than the OFDM system in a high-speed motion environment, and is more suitable for dual-selection channels. It can also be seen that when the OTFS system uses the ZF equalizer, if OPA is not performed, the performance is not as good as the OFDM system in the range of 0 to 20dB. This is also in line with the above-mentioned FD-ZF equalizer is more sensitive to the power allocation in the TF domain, while the OFDM system is not so sensitive to power allocation. However, it can be seen from the slope of the curve that the curve of the OFDM system is relatively straight, while the curve of the OTFS system is steeper, so the OTFS system has better potential.

[0085] 3.2 Comparison of BER-SNR curves of noise introduced by different IDs when OPA is used in this system:

[0086] Figure 3 The BER-SNR curves of the additional noise introduced by different IDs when applying the OPA scheme are compared. At this time, the number of channel multipaths is set to Q=5, where the noise power of different IDs is equal to the noise power brought by the channel, one-tenth of the channel noise power, and the ideal situation where the ID does not bring noise. It can be seen that compared with the FD-ZF equalizer, the FD-MMSE equalizer is more sensitive to the ID noise power. If the ID noise is relatively large, the gain attenuation of the FD-MMSE equalizer is relatively large. It can also be found that the performance of the FD-ZF equalizer is attenuated at about 20dB, and an error flattening is gradually produced, while FD-MMSE uses complexity to achieve better performance.

[0087] 3.3 Comparison of BER-SNR curves for different multipath numbers when OPA is applied in this system:

[0088] Figure 4 The BER-SNR curves of different channel multipath numbers when applying the OPA scheme are compared, where the additional noise power introduced by setting the ID is one-tenth of the channel noise power. It can be seen that as the number of paths increases, the OPA method can obtain a higher degree of freedom of selection, thereby obtaining a greater bit error rate gain. This is because the more drastic the channel fluctuations in the TF domain, the more obvious the pre-equalization effect of the OPA. In addition, it can be found that the FD-MMSE equalizer obtains a higher gain than the FD-ZF equalizer when the number of channels increases, at the cost of its higher complexity.

[0089] 4. Conclusion

[0090] The present invention designs the model of the OTFS-SWIPT system, derives its input-output relationship according to the mathematical expression, performs power allocation at the transmitting end, introduces the frequency domain linear equalizer at the receiving end, and constructs the optimization problem of power allocation to maximize the SNR or SSINR. The proposed system is simulated and numerically verified, and the results show that the system with the proposed OPA scheme has a lower BER than the system with the EPA scheme, which improves the reliability of communication.

Claims

1. An OTFS wireless data and energy simultaneous transmission method based on frequency domain equalizer in a high mobility environment, characterized in that: The process includes: Step 1: At the transmitting end, the input DD domain signal is transformed into the TF domain by inverse symplectic finite Fourier transform; Step 2, applying the power allocation matrix to the TF domain signal to obtain a signal after power allocation; Step 3, perform Heisenberg transform on the power-allocated signal to obtain a time domain signal, which is sent out through the channel; Step 4, at the receiving end, the received time domain signal is then power split; Step 5: The power-segmented signal is converted back to the TF domain through Wigner transformation, and energy is collected at the same time; Step 6: Perform frequency domain equalization on the TF domain signal, and then perform symplectic finite Fourier transform, that is, convert it back to DD domain signal.

2. According to the OTFS wireless data and energy simultaneous transmission method based on frequency domain equalizer in a high mobility environment in claim 1, it is characterized in that: Frequency domain equalization is processed using frequency domain linear equalizers, including ZF equalizer and MMSE equalizer.

3. The OTFS wireless data and energy simultaneous transmission method based on frequency domain equalizer in a high mobility environment according to claim 2 is characterized in that: When using the ZF equalizer for frequency domain equalization, the following optimization problem is established for the power allocation and power splitting parameters: in, In the formula, p nM+m are the diagonal elements in the power allocation matrix P, M and N are the number of subcarriers and time slots respectively, P is a diagonal matrix of size NM*NM, λ is the division factor of power division, H[n,m] is the channel response in the TF domain, is the energy conversion efficiency, e is the minimum power required for energy harvesting EH, σ is the variance of the sign, P T is the maximum transmission power allowed by the system, is the noise power received by the receiving antenna, σ 2 The additional noise power introduced when the receiving end decodes the information; Solving the optimization problem obtains the power allocation matrix P and the power splitting factor λ.

4. The OTFS wireless data and energy simultaneous transmission method based on frequency domain equalizer in a high mobility environment according to claim 2 is characterized in that: When using MMSE equalizer for frequency domain equalization, the following optimization problem is established for power allocation and power splitting parameters: in, In the formula, p nM+m are the diagonal elements in the power allocation matrix P, M and N are the number of subcarriers and time slots respectively, P is a diagonal matrix of size NM*NM, λ is the division factor of power division, H[n,m] is the channel response in the TF domain, is the energy conversion efficiency, e is the minimum power required for energy harvesting EH, ρ is the variance of each symbol, P T is the maximum transmission power allowed by the system, is the noise power received by the receiving antenna, σ 2 The additional noise power introduced when the receiving end decodes the information; Solving the optimization problem obtains the power allocation matrix P and the power splitting factor λ.

Images