KR20050005054A - Apparatus and Method for Generating Noise of Test Channel - Google Patents

Abstract

PURPOSE: A device and a method for generating noise of a test channel are provided to realize an AWGN(Additive White Gaussian Noise) model used for a virtual channel environment test as hardware in order to improve a data processing rate, thereby conducting the test in real time while improving reliability on test results. CONSTITUTION: A uniform random generator(110) generates a uniform random value for Gaussian distributions. A variable generator(120) generates a Gaussian variable by using the uniform random value. A noise generator(130) applies the Gaussian variable to the uniform random value, and generates Gaussian noise. The variable generator(120) comprises as follows. Squaring units(121,122) square the uniform random value to output the squared value. An adder(123) adds the square of the uniform random value to calculate a uniform deflection. A comparator(124) compares the uniform deflection with a Gaussian deflection threshold value, and outputs a smaller uniform deflection. A memory(125) prestores the Gaussian variable, and outputs the prestored Gaussian variable.

Description

Apparatus and Method for Generating Noise of Test Channel

The present invention relates to an apparatus and method for generating noise in a test channel, and in particular, to implement an additive white gauge model used in a virtual channel environment test in hardware to improve the test data processing speed. It is about.

In general, a channel environment test of wireless communication is performed in a virtual channel environment in consideration of time and cost.

In such a virtual channel environment, a model that generates constant noise in consideration of ambient noise is used, and a representative model is an AWGN model.

The AWGN model may be implemented by software or hardware. However, when implemented in software, many simulation timings are required during software operation. The operating frequency of the 3GPP baseband test is 3.84 MHz. If only one channel is tested, real-time processing is possible, but the channel test environment includes not only AWGN but also many other channel environments. If time is used, real-time processing will not be possible even if a high speed DSP (Digital Signal Processor) is used. In fact, using the TI DSP 320C6416 takes more than 100 times the real-time processing.

Therefore, for real-time processing in the 3GPP asynchronous wireless channel environment simulation, a hardware design that performs 3GPP baseband tests in real time is essential.

However, the conventional AWGN generation device has a problem of increasing the overall hardware size because many hardware paths are required to calculate the Gaussian parameters necessary for noise generation.

The present invention is to solve the problems as described above, the purpose is to implement the AWGN model used for the virtual channel environment test in hardware to improve the data processing speed to perform the test in real time, improve the reliability of the test results It is.

Furthermore, another object of the present invention is to reduce the hardware size by implementing a Gaussian variable calculation path as a predetermined memory in which a Gaussian variable corresponding to each uniform deviation that can occur in the Gaussian distribution is stored.

1 is a block diagram of an AWGN generating apparatus according to an embodiment of the present invention.

2 is a block diagram of an AWGN generating apparatus according to another embodiment of the present invention.

3 is a flow chart illustrating an AWGN generation procedure in FIG.

4 is a diagram illustrating a BER value for Eb / No in a virtual channel environment using the AWGN of the present invention.

Explanation of symbols on the main parts of the drawings

110,210: uniformly distributed value generator 120,220: variable generator

130,230: Noise Generator 121,122,221: Squarer

123,222 Adder 124,223 Comparator

125,224: First memory 131,132,231: Delay

133,134,232: multiplier 135,233: second memory

An apparatus for generating noise in a test channel of the present invention for achieving the above object includes: a uniform distribution value generator for generating a uniform distribution value on a Gaussian distribution; A variable generator for generating a Gaussian variable using the uniform distribution value; And a noise generator for generating Gaussian noise by applying the Gaussian variable to the uniform distribution.

Preferably, the variable generator is a squarer for outputting squared uniform distribution value; An adder for calculating a uniform deviation by adding a square of the uniform distribution value; A comparator for comparing the uniform deviation with a previously stored Gaussian deviation threshold and outputting a uniform deviation smaller than the threshold; And a memory which pre-stores a Gaussian variable corresponding to each uniform deviation that may occur, and outputs a Gaussian variable corresponding to the comparator output uniform deviation.

Also preferably, the noise generator may include a delay unit configured to delay the uniform distribution value for a predetermined time; A multiplier for generating the Gaussian noise by multiplying the delayed uniform distribution value and the Gaussian variable; And a memory for storing the Gaussian noise.

Furthermore, the Gaussian noise generating method according to another embodiment of the present invention includes the steps of generating a uniform distribution value on the Gaussian distribution; Generating a Gaussian variable using the uniform distribution value; And applying Gaussian parameters to the uniform distribution to generate Gaussian noise.

Preferably, the generating of the Gaussian variable comprises: square the uniform distribution value; Calculating a uniform deviation by adding a square of the uniform distribution value; Comparing the uniformity deviation with a previously stored Gaussian deviation threshold; And outputting a pre-stored Gaussian variable corresponding to a uniform deviation smaller than the threshold.

Also preferably, the generating of the Gaussian noise may include delaying the uniform distribution value by a predetermined time; Generating the Gaussian noise by multiplying the delayed uniform distribution value by the Gaussian variable; And storing the generated Gaussian noise.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an AWGN generating apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the AWGN generator according to the present embodiment includes a uniform random generator 110, a variable generator 120, and a noise generator 130. In addition, the variable generator 120 includes a first squarer 121, a second squarer 122, an adder 123, a comparator 124, and a first memory 125. A first delay unit 131, a second delay unit 132, a first multiplier 133, a second multiplier 134, and a second memory 135 are formed.

The uniform distribution value generator 110 generates a first uniform distribution value V1 and a second uniform distribution value V2 as a function generator having a uniform distribution.

The first squarer 121 and the second squarer 122 square the uniform distribution values V1 and V2 generated by the uniform distribution value generator 110, and the first squarer 121 has a first uniform distribution value V1. Squared, and the second multiplier 122 squares the second uniform distribution value V2. Since the first squarer 121 and the second squarer 122 have a large hardware size and a large propagation delay, a register is added to an output terminal to operate at a desired operating frequency (3.84 * 2 MHz). .

An adder 123 adding the output V2 of the ^second first squarer 121 outputs V1 ² and a second squarer 122, and outputs a uniform deviation R.

The comparator 124 compares the uniform deviation R output from the adder 123 with the Gaussian deviation threshold of 1, so that the final AWGN data has a Gaussian distribution, and discards the corresponding R if R> 1. In this case, the corresponding R is output together with the use signal of the first memory 125. At this time, since the comparator 124 operates at a speed of 3.84 * 2 MHz, an error that the data R for generating Gaussian noise is insufficient due to the comparison operation does not occur.

The first memory 125 is a component for solving a hardware size increase problem generated in a hardware implementation for generating AWGN data. In other words, the equation for calculating the Gaussian variable FAC has a large amount of calculation, which causes a hardware size increase in the hardware configuration, and also requires a pipeline technique for high-speed computation. High register usage also contributes to increased hardware size.

Accordingly, in order to solve this problem, the first memory 125 is implemented as a ROM in which a Gaussian variable FAC for the uniform deviation R is stored in a portion designated by the uniform deviation R value.

The first memory 125 has a label corresponding to a uniform deviation R that can be generated through simulation considering hardware size and bit error rate (BER), and each label is a result value of applying each uniform deviation R to the following equation. A Gaussian variable FAC is stored, and a Gaussian variable FAC stored in a corresponding label is output using the uniform deviation R output from the comparator 124 as an address.

FAC = sqrt (2.0 * -log (R) / R)

The first delay unit 131 and the second delay unit 132 are each implemented with a predetermined number of registers to delay the uniform distribution value generated by the uniform distribution value generator 110 by a predetermined time. Is delayed by receiving the first uniform distribution value V1, and the second delay unit 132 receives and delays the second uniform distribution value V2.

The first multiplier 133 multiplies the Gaussian variable FAC output from the first memory 125 by the uniform distribution value V1 delayed by the first delay unit 131 to output the first AWGN data gstore, and the second multiplier 134. Outputs the second AWGN data gaus by multiplying the Gaussian variable FAC output from the first memory 125 by the uniform distribution value V2 delayed by the second delay unit 132.

The first and second AWGN data are stored in the second memory 135 and used for the virtual channel test. At this time, the second memory 135 writes AWGN data at a rate of 3.84 * 2 MHz in one slot (2560chipx1) and AWGN data at a rate of 3.84MHz in another slot in order not to run out of AWGN data used for the test. Two slots of the 3GPP standard to read are used.

Each component of the AWGN generation device according to the present embodiment operates at a frequency of 3.84 * 2 MHz, which is a frequency for processing test data in real time, considering the probability that the uniform deviation R is less than 1. That is, the probability that the uniform deviation R is less than 1 in the 3GPP standard is 78.5%. Therefore, when each component operates as chipx1, only 78.5% of the test data can be processed and real-time processing is impossible, so the chipx2 is operated at a frequency of 3.84 * 2MHz.

2 is a block diagram of an AWGN generating apparatus according to another embodiment of the present invention.

Referring to FIG. 2, the AWGN generator of the present exemplary embodiment includes a uniform random generator 210, a variable generator 220, and a noise generator 230 as in FIG. 1. However, unlike FIG. 1, in order to reduce hardware size, the variable generator 220 includes a squarer 221, an adder 222, a comparator 223, and a first memory 224, and the noise generator 230 has a delay. Group 231, a multiplier 232, and a second memory 233. Each of the components operates at a frequency of 3.84 * 4 MHz in consideration of the hardware sharing in which the two components are processed separately in one component in FIG. 1.

The uniform distribution value generator 210 sequentially generates the first uniform distribution value V1 and the second uniform distribution value V2.

The squarer 221 squares and outputs the uniform distribution values V1 and V2 sequentially generated by the uniform distribution value generator 210, respectively.

The adder 222 adds the outputs V1 ² and V2 ² of the squarer 221 to output a uniform deviation R. The V1 ² is stored ^first and V2 ² is added to the stored V1 ² when V2 ^{2 is} input.

The comparator 223 compares the uniform deviation R output from the adder 222 with 1 so that the final AWGN data has a Gaussian distribution, and discards the corresponding R when R> 1, and removes the corresponding R when R <1. 1 Outputs with the memory 224 use signal. At this time, since the comparator 223 operates at a speed of 3.84 * 4 MHz, the error that the data R for AWGN generation is insufficient due to the comparison operation does not occur.

The first memory 224 is implemented as a ROM, and has a label corresponding to the number of branches of the uniform deviation R. Each label stores a Gaussian variable FAC, which is a result of applying each uniform deviation R to Equation 1, and a comparator. The Gaussian variable FAC stored in the label is output by using the uniform deviation R output from the (223) as an address.

The delay unit 231 is implemented with a predetermined number of registers to sequentially receive the outputs V1 and V2 of the uniform distribution value generator 210 and delay the predetermined time.

The multiplier 232 outputs the AWGN data by multiplying the Gaussian variable FAC, which is the output of the first memory 224, with the uniformly distributed values V1, V2 delayed by the delayer 231. Next, multiply FAC by V2 to output second AWGN data (gaus).

The first AWGN data and the second AWGN data are stored in the second memory 233 to be used for the real-time virtual channel test. At this time, the second memory 223 uses two slots of the 3GPP standard for writing AWGN data at a speed of 3.84 * 4 MHz in one slot (2560chipx1) and reading AWGN data at a speed of 3.84MHz in another slot.

3 is a flowchart illustrating an operation procedure of an AWGN generating apparatus according to the embodiment of FIG. 2.

Referring to FIG. 3, first, the uniform distribution value generator 210 sequentially generates uniform distribution values V1 and V2 (S31).

Then, the squarer 221 squares the sequentially generated V1 and V2, and the adder 222 adds V1 ² and V2 ² output from the squarer 221 to calculate the uniform deviation R, while the delayer 231 delays the sequentially generated V1 and V2 for a predetermined time (S32).

Thereafter, the comparator 223 checks whether the calculated R <1 (S33), and if R> 1, discards the corresponding R (S34). If R <1, the comparator 223 stores the corresponding R in the first memory 224. The first memory 224 outputs a Gaussian variable FAC stored at an address corresponding to the output R, together with a usage signal (S35).

The multiplier 232 multiplies the Gaussian variable FAC by the outputs V1 and V2 of the delayer 231 to generate first AWGN data (gstore = V1 * FAC) and second AWGN data (gaus = V2 * FAC). The data is stored in the second memory 233 (S36).

FIG. 4 is a diagram illustrating a BER measurement result of a virtual channel test using AWGN data generated by an AWGN generator according to the present invention according to a spread factor (SF) of a quadrature phase shift keying (QPSK) method. This is the result of BER measurement in a situation where the operating speed is satisfied in real time.

The BER measurement method generates an uplink transmission channel and a reception channel using a field programmable gate array (FPGA). In this case, the transmission channel transmits Dedicated Physical Control CHannel (DPCCH) data and Dedicated Physical Data CHannel (DPDCH) data at the same rate, and the reception channel receives only DPDCH data. The receiving end measures the BER of the received DPDCH using AGC (Automatic Gain Control).

Referring to FIG. 4, Q Eb / No (energy vs. noise spectral density of user bits) is a theoretical BER value when AWGN is added to DPDCH data, and SF4 Eb / No, SF16 Eb / No, and SF64 Eb / No. Are BER values of DPDCH data in the case of SF4, SF16, and SF64, respectively. As a result of performing a virtual test using the AWGN data generated by the AWGN generating apparatus of the present invention, it can be seen that the BER values for each SF are almost identical to the theoretical values.

In addition, the embodiment according to the present invention is not limited to the above description, and various alternatives, modifications, and changes can be made within the scope apparent to those skilled in the art.

As described above, the present invention implements the AWGN model used for the virtual channel environment test in hardware, thereby improving the data processing speed and performing the test in real time, further improving the reliability of the test results.

In addition, by implementing a Gaussian variable calculation path, which causes a size increase in hardware, as a predetermined memory in which Gaussian variables corresponding to the uniform deviations that may occur in the Gaussian distribution are implemented, the hardware size can be reduced.

Claims (6)

A uniform distribution value generator for generating a uniform distribution value on a Gaussian distribution; A variable generator for generating a Gaussian variable using the uniform distribution value; And a noise generator for generating Gaussian noise by applying the Gaussian variable to the uniform distribution value. The method of claim 1, The variable generator includes: a squarer for outputting squared uniform distribution values; An adder for calculating a uniform deviation by adding a square of the uniform distribution value; A comparator for comparing the uniform deviation with a previously stored Gaussian deviation threshold and outputting a uniform deviation smaller than the threshold; And a memory which pre-stores a Gaussian variable corresponding to each uniform deviation that may occur, and outputs a Gaussian variable corresponding to the comparator output uniform deviation. The method of claim 1, The noise generator includes a delay unit for delaying the uniform distribution value for a predetermined time; A multiplier for generating the Gaussian noise by multiplying the delayed uniform distribution value and the Gaussian variable; And a memory for storing the Gaussian noise. Generating a uniform distribution value on the Gaussian distribution; Generating a Gaussian variable using the uniform distribution value; And generating Gaussian noise by applying the Gaussian variable to the uniform distribution. The method of claim 4, wherein Generating the Gaussian variable includes: squaring the uniform distribution value; Calculating a uniform deviation by adding a square of the uniform distribution value; Comparing the uniformity deviation with a previously stored Gaussian deviation threshold; And outputting a pre-stored Gaussian variable corresponding to the uniform deviation smaller than the threshold value. The method of claim 4, wherein The generating of the Gaussian noise may include: delaying the uniform distribution value by a predetermined time; Generating the Gaussian noise by multiplying the delayed uniform distribution value by the Gaussian variable; And storing the generated Gaussian noise.