Building sound insulation performance detection method and multi-point automatic measurement system

Abstract

The application relates to the technical field of architectural acoustics detection, and discloses a building sound insulation performance detection method and a multi-point automatic measurement system. The method comprises the following steps: a master control end broadcasts a precise time protocol frame and plays a broadband pseudo-random sequence excitation signal in a sound source room; a sensing node in the room calibrates a clock, divides collected sound pressure signals into data frames, and compresses the data frames into a local first-in-first-out ring buffer; the sensing node performs cross-correlation operation on the data frames and the excitation signal to calculate a peak signal-to-noise ratio, and if the peak signal-to-noise ratio is lower than a threshold value, the data frame validity state is overwritten as invalid; after the node experiences a deterministic back-off time, a global veto beacon containing a serial number is broadcasted; after the remaining nodes receive the beacon, data frames corresponding to the local serial number are forcibly overwritten as invalid; each node performs integral calculation on the data frames maintaining the valid state and returns equivalent continuous sound pressure level data. The application reduces communication bandwidth occupation and improves the real-time performance of system detection.

Description

Technical Field

[0001] This invention relates to the field of building acoustic testing technology, specifically to a method for testing the sound insulation performance of buildings and a multi-point automatic measurement system. Background Technology

[0002] Building sound insulation performance testing is used to assess the acoustic environment quality inside buildings. The conventional measurement process involves playing acoustic excitation signals in the building's sound source room and deploying audio sensors in the receiving room to measure the transmitted sound pressure level. To address the need for simultaneous measurement in large spaces and at multiple points, existing sound insulation testing systems often employ a wireless sensor network architecture, deploying multiple wireless sensor nodes in the receiving room to replace manual relocation measurements or centralized multi-channel cable measurements.

[0003] In this distributed wireless measurement architecture, the system has shortcomings in handling acoustic data and sudden background noise. Sudden background noises such as door closing sounds, footsteps, or vehicle horns can easily be mixed in during the building site testing period. Existing wireless multi-point detection systems mostly adopt a centralized data processing mode. Sensor nodes continuously collect sound pressure signals and convert them into digital sequences, then directly transmit the complete digital signals back to the master control unit via a wireless radio frequency channel. After collecting data from all nodes, the master control unit uniformly performs post-processing operations such as cross-correlation analysis, signal-to-noise ratio calculation, and noise removal. Due to the large amount of acoustic data, the centralized transmission mechanism consumes a significant amount of communication bandwidth, and channel congestion and data loss are easily caused when multiple nodes operate concurrently.

[0004] Furthermore, when the master control unit detects sudden noise interference in the signal within a certain time window during the data analysis phase, it typically needs to discard all data transmitted from all nodes within that time period and notify each node to re-initiate the excitation and acquisition process. This approach increases the system's detection time in noisy environments. If each sensor node independently judges and discards interference data locally, the timing and amount of data discarded by each node will be inconsistent due to differences in the sound field at different node locations. This will cause the multi-point sound pressure level data aggregated by the master control unit to be misaligned on the time axis, making it difficult to meet the requirements of acoustic measurement specifications for spatial and temporal averaging of equivalent continuous sound pressure levels. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for testing the sound insulation performance of buildings and a multi-point automatic measurement system. This solves the technical problems of communication channel congestion caused by the centralized transmission of full sound pressure data and the time asynchrony of multi-point data caused by each node independently processing sudden noise in existing distributed wireless building sound insulation testing systems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a method for testing the sound insulation performance of a building, comprising the following steps: The master control unit broadcasts a precise time protocol frame and plays a broadband pseudo-random sequence excitation signal in the sound source room of the target building; multiple sensor nodes deployed in the receiving room of the target building receive the precise time protocol frame to calibrate the local real-time clock and establish a discrete time window sequence. The sensor node collects the sound pressure signal, converts it from analog to digital, divides it into data frames, assigns a sequence number and an initial valid status flag, and pushes it into the local first-in-first-out circular buffer. The sensing node performs cross-correlation calculation on the data frame and the broadband pseudo-random sequence excitation signal to calculate the peak signal-to-noise ratio. If the peak signal-to-noise ratio is lower than the preset threshold, the validity status flag bit of the data frame is overwritten to invalid status. When the validity status flag is overwritten to invalid, the sensor node calculates the deterministic backoff time, listens to the radio frequency channel after the deterministic backoff time, and broadcasts a global veto beacon with a sequence number when the radio frequency channel is idle. After receiving the global veto beacon, the remaining sensor nodes will forcibly overwrite the validity status flag of the data frame with the corresponding sequence number in their local first-in-first-out circular buffer to an invalid status. The sensor node performs logarithmic integral calculation of acoustic energy on the data frame that maintains an effective state, and transmits the equivalent continuous sound pressure level data back to the main control terminal when the preset test duration lower limit is reached.

[0007] Furthermore, the microcontroller of the sensor node records a local timestamp based on the hardware interrupt signal upon completion of RF reception. Combining this with the transmission timestamp in the precise time protocol frame and a fixed propagation delay constant, it calculates and compensates for the time offset to the local real-time clock. The microcontroller uses timer interrupts to divide physical time into discrete time windows. The microcontroller partitions contiguous memory blocks in memory as a first-in-first-out circular buffer, and the total number of slots determines the physical dwell time, which is strictly greater than the system's maximum total communication delay. This total communication delay encompasses the microcontroller's maximum computation time, maximum deterministic backoff time, RF over-the-air transmission time, and the processing time of the microcontrollers of the other sensor nodes.

[0008] Furthermore, the microcontroller calls the hardware floating-point unit to perform a fast Fourier transform, mapping the data frame to the frequency domain to obtain a complex spectrum, and performing point-by-point complex multiplication of array elements with the reference conjugate spectrum data of the broadband pseudo-random sequence excitation signal pre-stored in flash memory; the resulting cross power spectrum is mapped back to the time domain by a fast inverse discrete Fourier transform to generate a discrete cross-correlation sequence. The microcontroller extracts the discrete-time offset index corresponding to the point with the maximum absolute value in the sequence as the position of the correlation peak, and constructs a cyclic boundary asymmetric acoustic energy masking window combined with modulo operation. This masking window has a fixed pre-sampling point length on the left side of the peak position and a post-retention length determined by the nominal reverberation time of the receiving room on the right side. The microcontroller multiplies the acoustic energy masking window with the discrete cross-correlation sequence point by point, truncates the background noise outside the masking window, and obtains the effective room impulse response sequence. The microcontroller calculates the sum of squares of amplitudes inside the effective room impulse response sequence as the effective signal energy, and the sum of squares of amplitudes of the remaining part after subtracting the effective room impulse response sequence from the original discrete cross-correlation sequence as the background noise energy. The two energy values ​​are added to a minimum positive bias constant and then divided, and a logarithmic operation is performed to obtain the peak signal-to-noise ratio.

[0009] Furthermore, the sensor node maps its integer identifier to a backoff slot multiple to calculate the deterministic backoff time. After the deterministic backoff time, the microcontroller reads the received signal strength indicator register to evaluate the RF channel energy amplitude. If it is less than the channel busy / idle threshold, a global veto beacon is broadcast. The global veto beacon adopts a fixed frame structure without routing information and without variable-length extension fields. It sequentially includes the core data payload of physical layer preamble, network synchronization word, bearer node identifier and corrupted data frame sequence number, and cyclic redundancy check code. The remaining sensor nodes respond to hardware interrupts to extract the target sequence number from the beacon, calculate the sequence number offset by comparing the latest generated data frame sequence number with the target sequence number, and use the current write pointer index minus the sequence number offset, add the physical slot depth of the circular buffer, and perform a modulo operation to directly locate the absolute storage index of the target data frame in physical memory, forcibly overwriting its validity status flag to invalid.

[0010] Furthermore, while maintaining the pop of valid state data frames from the stack, the microcontroller maintains the valid signal energy accumulator and the valid data frame count statistic in parallel. When the total number reaches the target valid frame count threshold, the acquisition stops. The equivalent time-averaged sound energy is obtained by dividing the total accumulated effective transmitted sound energy by the total number of valid data frames participating in the integration, and is then sent back to the main control terminal as equivalent continuous sound pressure level data.

[0011] A second aspect of the present invention provides a multi-point automatic measurement system for building sound insulation, used to implement the aforementioned method for testing building sound insulation performance, comprising: The main control unit is deployed inside the sound source room of the target building, and the receiving node array is deployed inside the receiving room of the target building; The main control unit includes an industrial computer, power amplifier, speaker, and main communication module; The industrial control computer is connected to the power amplifier, the power amplifier is electrically connected to the speaker, and the main communication module is connected to the industrial control computer. The receiving node array includes multiple sensing nodes configured with globally unique integer identifiers; The sensing node includes a microphone, an analog-to-digital converter, a microcontroller, a memory, and a node communication module; The microphone is connected to the analog-to-digital converter, which is connected to the microcontroller. The microcontroller has a built-in hardware floating-point unit and is connected to the internal first-in-first-out circular buffer and node communication module. The node communication module is used for wireless radio frequency communication with the main communication module and the node communication modules of other sensor nodes.

[0012] This invention provides a method for testing the sound insulation performance of buildings and a multi-point automatic measurement system. It has the following beneficial effects: 1. This invention effectively reduces the communication bandwidth consumption of wireless radio frequency networks and improves system real-time performance. By configuring a FIFO circular buffer locally on each sensor node and managing it in a distributed manner using the sequence number of the data frame, when a node detects noise contamination in its local signal, it only needs to broadcast a global veto beacon containing that sequence number. After receiving the beacon, the other nodes directly locate and overwrite the state of the corresponding data frame in their local memory using the sequence number offset. This mechanism avoids centrally transmitting massive amounts of raw audio data back to the master control unit for unified judgment, and directly completes the global removal of invalid data at the edge nodes.

[0013] 2. This invention improves the accuracy of bottom-level pulse signal extraction and signal-to-noise ratio (SNR) calculation in building sound insulation testing. After acquiring the discrete cross-correlation sequence, the microcontroller of the sensing node constructs an asymmetric acoustic masking window based on the position of the main correlation peak. The post-retention length on the right side of this masking window is determined by the nominal reverberation time of the target building's receiving room, enabling precise truncation and separation of the effective room impulse response sequence from the background noise. By calculating the peak SNR through this masking extraction method that combines actual acoustic environment physical parameters, interference from sudden noises in complex environments is eliminated, ensuring the reliability of subsequent data frames involved in acoustic energy integration.

[0014] 3. This invention solves the radio frequency channel conflict problem in multi-node concurrent communication, ensuring low-latency transmission of control commands. Before triggering a global veto broadcast, each node maps its own integer identifier to a multiple of the backoff time slot to obtain a deterministic backoff time, and after the backoff ends, it uses a channel energy amplitude listening mechanism to determine whether to send. When multiple sensor nodes are simultaneously subjected to sudden noise interference and determine that the data frame is invalid, this mechanism can use the uniqueness of the physical identifier to stagger the beacon transmission timing of each node, and combined with a fixed-length frame structure without variable-length extension fields, ensures a high success rate of beacon transmission between node arrays. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the system architecture of the present invention; Figure 3 The figures show the experimental verification and effect comparison of the building sound insulation measurement system of the present invention under sudden noise interference, where (a) is a comparison curve of the measurement error changing with the frequency of sudden noise occurrence; and (b) is a comparison curve of the cumulative data transmission volume of the wireless network changing with time. Detailed Implementation

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see the appendix Figure 2 This invention provides a multi-point automatic measurement system for building sound insulation, which may include: a main control terminal and a receiving node array.

[0018] The main control unit is deployed inside the sound source room of the target building. The main control unit includes an industrial computer, power amplifier, speakers, and main communication module.

[0019] The industrial computer is connected to the power amplifier. The power amplifier is electrically connected to the speaker. The industrial computer is used to generate a broadband pseudo-random sequence excitation signal. The industrial computer sends the broadband pseudo-random sequence excitation signal to the power amplifier for signal amplification.

[0020] The power amplifier drives the loudspeakers to radiate a sound field into the space within the sound source room. The main communication module is connected to the industrial control computer. The main communication module is used to broadcast time synchronization frames to the receiving node array and to receive spatial acoustic measurement data from the receiving node array.

[0021] The receiver node array is deployed inside the receiver room of the target building. The receiver node array consists of multiple sensor nodes. Each sensor node is configured with a globally unique integer identifier.

[0022] The sensing node includes a microphone, an analog-to-digital converter, a microcontroller, a memory, and a node communication module.

[0023] The microphone is used to acquire and receive the spatial sound pressure signal within the room. The microphone is connected to an analog-to-digital converter (ADC). The ADC converts the analog spatial sound pressure signal into a digital audio stream, which is then input to the microcontroller.

[0024] The microcontroller has a built-in hardware floating-point unit. It connects to the memory and the node communication module. The microcontroller utilizes the physical memory space of the memory to construct a first-in-first-out circular buffer.

[0025] The node communication module is used for radio frequency communication with the main communication module and the node communication modules of other sensor nodes. A low-power wide-area network (LPWAN) protocol is used to establish a wireless communication link between the main communication module and the node communication modules.

[0026] Please see the appendix Figure 1 This invention provides a method for testing the sound insulation performance of buildings, comprising the following steps: S1, the main control unit broadcasts a precise time protocol frame through the main communication module, the sensor node receives the precise time protocol frame to calibrate the microcontroller's local real-time clock, the system establishes a unified discrete time window sequence, and the main control unit drives the loudspeaker to play a broadband pseudo-random sequence excitation signal in a loop in the sound source room. S2, the sensor node continuously collects the spatial sound pressure signal of the receiving room through the microphone, the microcontroller divides the digital audio stream into data frames according to the time window length, the microcontroller pushes the data frames into the first-in-first-out circular buffer inside the memory for delayed residence, and the microcontroller assigns an incrementing sequence number and an initial validity status flag bit to each data frame. S3, the microcontroller calls the hardware floating-point unit to perform cross-correlation operation on the currently resident data frame and the broadband pseudo-random sequence excitation signal. The microcontroller constructs a noise floor extraction model including an acoustic masking window and calculates the peak signal-to-noise ratio of the current data frame. The microcontroller compares the peak signal-to-noise ratio with a preset threshold. When the peak signal-to-noise ratio is lower than the preset threshold, the microcontroller overwrites the validity status flag of the current data frame to invalid. S4. When the sensor node triggers the data frame status bit overwrite operation, the sensor node extracts its own integer identifier to calculate the deterministic backoff time. After the backoff ends, the sensor node listens to the radio frequency channel through the node communication module. If the radio frequency channel is in an idle state, the sensor node broadcasts a global veto beacon containing the node identifier and the corrupted data frame sequence number. If a beacon carrying the same sequence number is detected, the sensor node cancels the broadcast operation. S5, the other sensor nodes in the receiving room that have not triggered the status bit overwrite operation receive the global veto beacon. The sensor node parses the sequence number in the global veto beacon and locates the data frame with the same sequence number in its own first-in-first-out circular buffer. The sensor node forcibly overwrites the validity status flag bit of the located data frame to invalid status. S6, when a data frame arrives at the exit of the first-in-first-out circular buffer, the microcontroller reads the validity status flag of the data frame. The microcontroller discards the data frame marked as invalid and extracts the data frame marked as valid to perform logarithmic integration of acoustic energy. When the total effective integration time of the sensor node reaches the preset test time lower limit, the sensor node sends back the equivalent continuous sound pressure level data to the master control terminal. The master control terminal calculates the spatial average sound pressure level of the receiving room based on the data sent back by all sensor nodes.

[0027] The precise time alignment method provided by this invention includes: The master control unit acquires the reference system time. The industrial control computer reads its internal high-precision clock chip as the global reference time. The industrial control computer encapsulates the global reference time into the data payload of a precise time protocol frame. The main communication module inserts a timestamp of the transmission time into the frame header of the precise time protocol frame. The precise time protocol frame is used to establish a unified time reference in the distributed network to ensure that acoustic acquisition actions distributed in different physical locations are strictly aligned on the time axis, thereby avoiding time deviations that could cause subsequent cross-correlation algorithms to fail.

[0028] The master control unit sends out precise time protocol frames. The master control unit broadcasts these frames throughout the target building space via the main communication module. A low-power wide-area network (LPWAN) protocol is used between the main communication module and the node communication modules. Specifically, the LPWAN protocol is either a wireless radio frequency communication protocol based on linear frequency modulation (LFM) spread spectrum modulation or an ultra-narrowband physical layer communication protocol. LFM spread spectrum modulation can provide long-distance radio frequency links in building environments with high physical barriers.

[0029] The sensor node receives the precise time protocol frame and calculates the time offset. The sensor node receives the precise time protocol frame through the node communication module. The microcontroller records the local timestamp of the arrival time based on the RF reception completion hardware interrupt signal generated by the node communication module. The microcontroller parses the precise time protocol frame to obtain the transmission timestamp recorded by the master control terminal. The microcontroller calculates the time offset between the local clock and the global reference time.

[0030] The formula for calculating the time offset is as follows: ; In the formula, For nodes The time offset that needs to be compensated. For nodes Generate a local timestamp when the RF reception hardware interrupt is completed. The transmission time timestamp is encapsulated in the precise time protocol frame by the main control unit. Let be the fixed propagation delay constant of the radio frequency signal in the air. This is an integer identifier for the sensor node. Wherein, the fixed propagation delay constant... The physical layer parameters, such as the spreading factor, coding rate, and signal bandwidth, of the main communication module and the node communication module are calculated in advance or obtained through two-way ranging calibration in a laboratory environment. The values ​​are usually in the range of milliseconds.

[0031] For the physical layer implementation principle of linear frequency modulation spread spectrum modulation and the method for determining the fixed propagation delay constant, those skilled in the art can consult relevant communication radio frequency technology manuals, which are well-known technologies in the field and will not be elaborated here.

[0032] The microcontroller calibrates the local real-time clock. The local real-time clock is provided by a physical clock source, which is provided by the microcontroller's internal hardware timer in conjunction with an external crystal oscillator. The microcontroller compensates for the calculated time offset in the local real-time clock register.

[0033] The calibration formula for a local clock is as follows: ; In the formula, For nodes The local real-time clock value after compensation and calibration. For nodes The value of the local real-time clock register before calibration. For nodes The time offset that needs to be compensated.

[0034] The microcontroller establishes a unified sequence of discrete time windows. After calibration, the local real-time clocks of all sensor nodes remain synchronized. The microcontroller continuously generates internal timing interrupts according to the preset time window length. The microcontroller divides continuous physical time into discrete time windows of equal length.

[0035] The formula for generating discrete time window sequences is as follows: ; In the formula, For the first The starting absolute physical time of a discrete time window The system agrees on the absolute start time for initiating spatial sound pressure acquisition. This time is specified by the master control terminal as a fixed future time point in the precise time protocol frame. This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This is the preset time window length. The number of acoustic sampling points contained in a single data frame is determined, and its value is typically set between 50 milliseconds and 200 milliseconds. The choice of this length requires balancing the frequency resolution of the cross-correlation algorithm with the RF communication latency tolerance of the system node communication modules. By generating a discrete time window sequence, the system maps independent acoustic acquisition actions scattered throughout the physical space onto a globally aligned data frame timeline.

[0036] Following the precise time alignment method provided by this invention, the system executes a deterministic stimulus construction mechanism, specifically including: The industrial control computer generates a preset broadband pseudo-random sequence excitation signal. This broadband pseudo-random sequence excitation signal is a discrete digital audio sequence with a deterministic distribution in the time domain. Conventional white noise or pink noise are non-stationary random signals, and their phases are disordered within short time segments, leading to fluctuations in short-time correlation calculations. In contrast, the broadband pseudo-random sequence excitation signal possesses strict periodicity and deterministic phase information, providing a stable correlation matching benchmark for the receiver.

[0037] In the specific implementation of the lower-level features, the broadband pseudo-random sequence excitation signal is either a maximum-length sequence or a logarithmically swept sinusoidal signal. The maximum-length sequence is generated through a multi-stage linear feedback shift register, and its frequency components uniformly cover the measurement frequency band. The instantaneous frequency of the logarithmically swept sinusoidal signal increases exponentially with physical time. The logarithmically swept sinusoidal signal can physically separate the linear response of the acoustic system from nonlinear harmonic distortion in the time domain, improving the system's immunity to nonlinear distortion under high sound pressure levels.

[0038] Industrial control computers utilize the autocorrelation characteristics of broadband pseudo-random sequence excitation signals to establish an anti-interference foundation. The ideal autocorrelation function of a broadband pseudo-random sequence excitation signal exhibits physical characteristics approximating the Dirac function. This physical characteristic ensures that when a digital audio sequence is cross-correlated with itself, a high energy peak is generated only at the zero time offset, while a low noise floor is observed at other time offsets.

[0039] In acoustic measurement principles, the sound signal received by the sensing node in the receiving room includes the excitation signal transmitted from the sound source room, reflected sound from the room, and external environmental noise. Since the external environmental noise and the broadband pseudo-random sequence excitation signal are uncorrelated, the energy generated by the environmental noise is evenly distributed in the underlying background during cross-correlation calculations, without producing correlation peaks. The received excitation signal, however, matches the original signal, forming a distinct energy peak. This physical characteristic allows the system to separate the deterministic test sound pressure from non-stationary interference noise at the algorithmic level, providing a basis for subsequent data truncation.

[0040] The formula for calculating the autocorrelation function of the broadband pseudo-random sequence excitation signal is as follows: ; In the formula, Let be the autocorrelation function of the broadband pseudo-random sequence excitation signal. This refers to the continuous time offset in autocorrelation calculations. Let be the cycle length of a single cycle of the broadband pseudo-random sequence excitation signal. For continuous physical time variables, For broadband pseudo-random sequence excitation signals in physical time Instantaneous amplitude at that point, The time offset of the broadband pseudo-random sequence excitation signal The instantaneous amplitude after the delay, This represents the total acoustic energy of the broadband pseudo-random excitation signal within a single cycle. For the Dirac function, For physical time The differential element (or time integral differential element).

[0041] The cycle length of a single cycle of a broadband pseudo-random sequence excitation signal The value is greater than the nominal reverberation time of the target building's receiving room. To prevent time aliasing in acoustic measurements—that is, the superposition of late-reflection acoustic energy from the previous cycle into the initial stage of the current cycle—the length of a single cycle is [not specified]. The value range is set to 2 seconds to 5 seconds.

[0042] The main control unit executes the cyclic playback of the broadband pseudo-random sequence excitation signal. The industrial control computer converts the generated digital audio sequence into an analog audio electrical signal. The industrial control computer outputs the analog audio electrical signal to the power amplifier. The power amplifier amplifies the voltage and current of the analog audio electrical signal. The power amplifier drives the loudspeakers to convert the amplified electrical signal into a spatial radiated sound field in the sound source room. The industrial control computer controls the broadband pseudo-random sequence excitation signal to play according to the length of a single cycle. Seamlessly looping and splicing playback is performed to form a continuous test sound field. The total loop playback duration set by the industrial control computer covers the minimum acoustic measurement time required by the system.

[0043] For the specific circuit topology of the digital-to-analog converter circuit inside the industrial control computer and the impedance matching amplification principle of the power amplifier, those skilled in the art can consult relevant electronic circuit design manuals, which are well-known technologies in the field and will not be elaborated here.

[0044] The specific implementation method of continuous audio acquisition and discrete time window mapping in the building sound insulation performance testing method provided by this invention is as follows: The sensor node acquires real-time spatial sound pressure signals from the receiving room via a microphone. The microphone converts the received continuous spatial sound pressure signals into analog voltage signals. The microphone then transmits the analog voltage signals to an analog-to-digital converter.

[0045] The analog-to-digital converter (ADC) discretizes the analog voltage signal according to a fixed system sampling rate. In the specific implementation, the ADC uses a 24-bit resolution ADC chip. The system sampling rate is set to 48 kHz to meet the Nyquist sampling theorem requirement for covering the core frequency band of 50 Hz to 5 kHz in architectural acoustics measurements. The ADC converts the continuous analog voltage signal into a discrete digital audio stream.

[0046] The microcontroller receives the digital audio stream and performs discrete-time window data partitioning. In the system's measurement principle, to ensure strict time-axis alignment of data collected by each sensor node in the distributed receiving node array, the microcontroller must rely on the absolute physical moment determined during the time synchronization phase as the truncation starting point for the first data frame. Without a globally unified starting point constraint, even slight differences in physical startup delays between different sensor nodes when activating the analog-to-digital converter will cause misalignment of the generated data frames globally, thus depriving the subsequent multi-node collaborative rejection mechanism of its comparison benchmark. Based on the previously established unified discrete-time window sequence, the microcontroller uses internal high-precision timer hardware interrupts to perform time-axis truncation on the continuous digital audio stream. The microcontroller divides the truncated digital audio stream into multiple fixed-length data frames.

[0047] The number of discrete sampling points contained in each data frame is determined by the time window length and the system sampling rate, and the calculation formula is as follows: ; In the formula, This represents the total number of discrete sampling points contained within a single data frame. The preset time window length, The system sampling rate set for the analog-to-digital converter. Combined with the preset time window length in the aforementioned embodiments. The range of values ​​for which a single data frame contains discrete sampling points. Typically in the thousands, this data length directly matches the single-processing capacity of the microcontroller's internal hardware floating-point arithmetic unit, avoiding memory overflow in the static random access memory due to excessively large single-frame data.

[0048] The microcontroller assigns a globally incrementing sequence number to each data frame. This sequence number uniquely identifies the absolute physical time segment corresponding to that data frame within the distributed receiver node array. The microcontroller then extracts the digital audio stream within a specified time window into independent discrete sequences.

[0049] The mathematical expression for frame mapping of a digital audio stream is as follows: ; In the formula, For sensing nodes In the serial number The first data frame A discrete digital sample amplitude, For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This is the internal index number of the discrete sampling points within the data frame, with a value ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. For sensing nodes The amplitude function of the acquired continuous digital audio stream, For the first The starting absolute physical time of a discrete time window The system sampling rate set for the analog-to-digital converter.

[0050] Through the aforementioned mapping mechanism, continuous digital audio streams are converted into data frames with global sequence numbers and fixed data lengths. These data frames serve as the basic processing units for subsequent underlying buffer delay and cross-correlation operations, ensuring the consistency of the data structure for acoustic data at different physical locations within the distributed receiving node array.

[0051] For the design of the anti-aliasing filter circuit inside the analog-to-digital converter and the underlying hardware principle of the microcontroller using the direct memory access controller to transmit audio data, those skilled in the art can consult relevant digital signal processing design manuals, as these are well-known technologies in the field and will not be elaborated here.

[0052] The specific implementation method of memory allocation and physical dwell time sequence of the annular buffer in the building sound insulation performance testing method provided by the present invention is as follows: After acquiring the segmented data frames, the sensor nodes need to buffer them. In a multi-node distributed acoustic measurement system, wireless channel communication inherently has physical delays. If a sensor node performs acoustic energy integration calculations immediately after acquiring a data frame, and one sensor node detects localized burst noise, the delay in radio frequency communication will prevent other sensor nodes from receiving warning information in time, thus mixing acoustically contaminated data frames into the global measurement results. To eliminate the asynchronous data processing among nodes caused by wireless communication delays, the microcontroller uses memory to establish a delay buffering mechanism.

[0053] The microcontroller allocates a contiguous block of memory within the physical memory space of the memory. This contiguous block is configured as a first-in, first-out (FIFO) circular buffer. The underlying data structure of the FIFO circular buffer includes a fixed total number of slots, a write pointer for pushing new data, and a read pointer for removing old data. When the write pointer reaches the end address of the contiguous memory block, the microcontroller controls the write pointer to wrap back to the beginning address of the contiguous memory block, thus forming a circular overwrite mechanism.

[0054] In the specific implementation of lower-level features, the microcontroller sets the total number of slots in the first-in-first-out circular buffer to the maximum communication delay tolerance of the system. The storage space allocated to each slot is equal to the product of the total number of discrete sampling points in a single data frame and the data bit width of a single sampling point, with an additional byte of memory allocated for storing the validity status flag bit of the data frame. Total number of slots The value is usually set to a positive integer between 50 and 200. The specific setting of this value is based on ensuring that the total dwell time of the data in the buffer is greater than the sum of the backoff delay and the radio frequency transmission time in the worst case of the wireless network, so as to ensure that the contaminated data frame has not entered the final integration accumulation unit before the rejection instruction arrives.

[0055] The microcontroller pushes a data frame with a globally incrementing sequence number into the slot pointed to by the current write pointer of the first-in-first-out (FIFO) circular buffer. Upon completion of the push operation, the microcontroller initializes the validity status flags of the data frame.

[0056] The initialization expression for the validity status flag is as follows: ; In the formula, For sensing nodes The collected data is in the serial number. The validity status flag bits of the data frame. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, with a value that is a non-negative integer. This indicates that the initial state of the data frame is valid. A valid state means that the data frame has not yet been determined to be contaminated by incoherent transient noise. If the system subsequently determines that the data frame is contaminated, its validity status flag will be overwritten with a value of 0, representing an invalid state.

[0057] After a data frame is pushed into the FIFO circular buffer, it does not immediately enter the final acoustic energy integration calculation stage, but remains resident in memory. Over time, the write pointer continuously stores newly generated data frames, and the read pointer removes older data frames from the other end of the FIFO circular buffer at equal time intervals. The physical time from when a data frame is pushed into the FIFO circular buffer to when it is removed from the FIFO circular buffer is defined as the physical residence time of the data frame.

[0058] The formula for calculating the physical dwell time of a data frame is as follows: ; In the formula, The physical residence time of a data frame in the first-in-first-out circular buffer. This represents the total number of slots in the first-in-first-out (FIFO) circular buffer. The preset time window length.

[0059] By setting the total number of slots The system reserves a defined physical dwell time for determining the validity of data frames. During this physical dwell time, the microcontroller executes a local anti-interference algorithm while the receiving node communication module obtains a global rejection beacon from the radio frequency channel. This time window reservation mechanism, from the perspective of hardware memory scheduling, ensures that the receiving node array has sufficient time to complete the spatial collaborative rejection determination among multiple sensor nodes, avoiding erroneous data integration caused by communication latency.

[0060] For the specific addressing logic of the write pointer and read pointer in the first-in-first-out circular buffer, as well as the direct memory access and transfer configuration of the memory, those skilled in the art can consult the relevant embedded system software programming manuals, which are well-known technologies in the field and will not be elaborated here.

[0061] The specific implementation of the constraint relationship between buffer depth and system communication delay in the building sound insulation performance testing method provided by this invention is as follows: In a multi-node distributed acoustic measurement system, when a sensor node detects sudden noise locally and triggers a spatial cooperative veto mechanism, the wireless communication link inherently experiences physical latency. This latency is mainly composed of local hardware computation time, channel backoff waiting time, radio frequency over-the-air transmission time, and remote interrupt handling time. During this latency, contaminated data frames within the remote sensor node continue to move towards the exit direction within the first-in-first-out circular buffer.

[0062] In the underlying operating mechanism of the system, there exists a physical time race: on one hand, the noise-contaminated data frame gradually approaches the final acoustic energy integration accumulation unit in memory as the pointer advances; on the other hand, the global veto beacon containing the interception command is being transmitted over the radio frequency channel. Setting the buffer depth essentially increases the physical residence time of the data frame in local memory, ensuring that the global veto beacon arrives at the exit point before the data frame in this time race. To ensure that the global veto beacon is received and processed by the remote sensor node before the contaminated data frame reaches the exit point of the FIFO circular buffer, the FIFO circular buffer configured by the microcontroller must meet a preset depth constraint. The system establishes a mathematical inequality relationship between the physical residence time of the FIFO circular buffer and the maximum total communication delay of the system.

[0063] The formula for calculating the maximum total communication delay of the system is as follows: ; In the formula, This represents the system's maximum total communication delay. The maximum computation time required for the microcontroller to call the hardware floating-point unit to perform cross-correlation calculations and peak extraction for a single frame of data. This represents the maximum deterministic backoff time allocated to the sensor node under carrier sense multiple access (CSMI) mechanism. To globally negate the radio frequency over-the-air transmission time of beacons at the physical layer of low-power wide-area networks, The processing time of the microprocessor in responding to the radio frequency receive interrupt and completing the internal address overwrite operation for the microcontroller of the remote sensing node.

[0064] In the implementation of specific lower-level features, the maximum computation time is... The maximum deterministic backoff time is determined by the microcontroller's clock frequency and the instruction cycle of the Fast Fourier Transform, and typically ranges from 5 to 10 milliseconds. The setting relies on the built-in integer identifier of the sensor node. The system assigns a unique backoff slot multiple to each sensor node. The maximum deterministic backoff time is obtained by multiplying the total number of sensor nodes in the distributed receiver node array by the basic backoff slot length, and its value is typically set between 50 milliseconds and 200 milliseconds. Radio frequency air transmission time. The time taken is calculated from the data payload length of the global veto beacon, the spreading factor, and the communication bandwidth, with typical values ​​between 20 and 50 milliseconds. Microprocessor processing time. It falls within the microsecond range, and in engineering calculations, it is fixed at 1 millisecond as a system tolerance margin.

[0065] Based on the maximum total system communication delay determined by the above parameters, the microcontroller allocates the total number of slots in the first-in-first-out (FIFO) circular buffer. The total number of slots constitutes the buffer depth, which, along with the maximum total system communication delay, must satisfy the following constraint inequality, thus deriving the total number of slots. Minimum value boundary: ; ; In the formula, This represents the total number of slots in the first-in-first-out (FIFO) circular buffer, and its value is a positive integer. The preset time window length, This represents the maximum computation time of the microcontroller. This represents the maximum deterministic backoff time of the sensor node. To globally reject the radio frequency over-the-air transmission time of the beacon, The processing time for the microprocessor of the remote sensing node.

[0066] When initializing the memory, the microcontroller sets the total number of slots. The value is set to the smallest integer greater than the ratio calculated on the right side above, with an additional 10% to 20% redundant slots. These redundant slots are used to absorb task scheduling jitter from the microcontroller's underlying operating system. By establishing these constraints, the system eliminates timing errors caused by RF communication delays affecting the effectiveness of acoustic measurement integration during the underlying memory allocation phase.

[0067] For the specific state machine implementation of the task scheduling mechanism of the real-time operating system inside the microcontroller and the channel medium access control protocol, those skilled in the art can refer to the relevant IoT communication protocol standards, which are well-known technologies in the field and will not be elaborated here.

[0068] The specific implementation method of dimensionality reduction processing for local hardware cross-correlation calculation in the building sound insulation performance testing method provided by this invention is as follows: In acoustic measurement principles, the microcontroller needs to calculate the cross-correlation sequence between the currently residing data frame and the original broadband pseudo-random sequence excitation signal to extract the correlation peak representing the effective acoustic energy. If discrete cross-correlation is performed directly in the time domain, its computational complexity is proportional to the square of the data frame length. Limited by the clock frequency and processing performance of the microcontroller's internal hardware floating-point unit, direct time-domain calculation would cause the processing time for a single frame to exceed the preset time window length, resulting in real-time data overflow and frame loss. To meet the system's underlying real-time requirements, the microcontroller uses frequency domain mapping to reduce the dimensionality of the cross-correlation operation. Simultaneously, since the industrial control computer cyclically plays broadband pseudo-random sequence excitation signals in space, this periodic physical sound field allows the microcontroller to directly use cyclic correlation based on discrete Fourier transform to equivalently perform the real linear cross-correlation operation without additional zero-padding of the data frame, thus maximizing the utilization of the microcontroller's memory space.

[0069] The specific dimensionality reduction process for cross-correlation operations includes the following steps: The microcontroller reads the pre-stored conjugate spectrum. The broadband pseudo-random sequence excitation signal has global determinism, and its complete discrete sequence is known before system operation. During the initialization phase, the system performs a discrete Fourier transform on the broadband pseudo-random sequence excitation signal in advance and obtains the complex conjugate to generate reference conjugate spectrum data. The microcontroller stores this reference conjugate spectrum data in its internal non-volatile flash memory. This implementation avoids the microcontroller repeatedly calculating the spectrum of the excitation signal in each loop, reducing the physical memory footprint of the static random access memory.

[0070] The microcontroller performs time-frequency conversion of the current data frame. The microcontroller calls the hardware floating-point unit and uses the Fast Fourier Transform algorithm to map the currently resident data frame in the first-in-first-out circular buffer from the time domain to the frequency domain, obtaining the complex spectrum of the current data frame.

[0071] The formula for time-frequency conversion is as follows: ; In the formula, For sensing nodes In the serial number The data frame has a frequency domain index number of Complex spectral values ​​at that location, For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This represents the total number of discrete sampling points contained within a single data frame. This is the internal index number of the discrete sampling points within the data frame, with a value ranging from 0 to... integers, For sensing nodes In the serial number The first data frame A discrete digital sample amplitude, This is the discrete frequency index number of the frequency domain sequence, with a value ranging from 0 to... integers, is the base of the natural logarithm. It is the imaginary unit.

[0072] The microcontroller performs point-to-point multiplication in the frequency domain. According to the correlation theorem in signal processing, the cross-correlation sequence of two time-domain signals is equivalent to multiplying their corresponding complex spectra by the conjugate complex spectrum of the other signal and then performing an inverse discrete Fourier transform. The microcontroller then performs point-by-point complex multiplication of the transformed complex spectrum of the current data frame with the conjugate spectrum of a pre-stored broadband pseudo-random sequence excitation signal to obtain the cross power spectrum.

[0073] The formula for calculating the cross power spectrum is as follows: ; In the formula, Frequency domain index number The complex value of the cross power spectrum at the location. The frequency domain index number of the current data frame Complex spectral values ​​at that location, The frequency domain index number of the pre-stored broadband pseudo-random sequence excitation signal. The conjugate complex spectrum value at that point. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This is the discrete frequency index number of the frequency domain sequence, with a value ranging from 0 to... Integers.

[0074] The microcontroller performs a reverse mapping to obtain the cross-correlation sequence. It performs a fast inverse discrete Fourier transform on the calculated cross-power spectrum, remapping the frequency domain multiplication result back to the time domain to obtain the dimension-reduced cross-correlation sequence.

[0075] The formula for calculating the inverse mapping is as follows: ; In the formula, To calculate the discrete cross-correlation sequence between the acquired current data frame and the broadband pseudo-random sequence excitation signal, This is the discrete-time offset index in the cross-correlation sequence, with a value ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. Frequency domain index number The complex value of the cross power spectrum at the location. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This is the discrete frequency index number of the frequency domain sequence, with a value ranging from 0 to... integers, is the base of the natural logarithm. It is the imaginary unit.

[0076] Through the aforementioned frequency domain mapping mechanism, the microcontroller reduces the complexity of local hardware cross-correlation operations from the quadratic level to the linear logarithmic level, ensuring that the processing time of data frames is strictly less than the physical acquisition time of a single audio frame, thus supporting the continuous and uninterrupted measurement of the system.

[0077] For the radix-2 butterfly graph operation bit-inverting addressing algorithm used in the fast Fourier transform inside the microcontroller, those skilled in the art can consult relevant digital signal processing algorithm design materials, which are well-known technologies in the field and will not be elaborated here.

[0078] The specific implementation method of dynamically constructing the room impulse response acoustic energy masking window in the building sound insulation performance testing method provided by the present invention is as follows: In acoustic measurement principles, the discrete cross-correlation sequence obtained by the microcontroller through dimensionality reduction calculation contains two parts of information. One part is the room impulse response reflecting the sound transmission path from the source room to the receiving room, which is represented by the main peak of energy concentration and the subsequent reverberation decay sequence; the other part is the background noise generated by environmental noise. In order to calculate the transmitted sound energy and eliminate environmental noise interference, the system needs to extract the time segment where the energy is effective. Based on the position of the correlation main peak, the microcontroller dynamically generates an asymmetric sound energy masking window in the time domain.

[0079] The microcontroller iterates through the discrete cross-correlation sequence, extracts the discrete-time offset index corresponding to the amplitude point with the largest absolute value, and uses this as the position of the main correlation peak. The position of the main correlation peak represents the arrival time of the directly transmitted sound wave at the microphone.

[0080] The formula for extracting the position of the main correlation peak is as follows: ; In the formula, This is the discrete-time offset index corresponding to the main correlation peak. To find the mathematical operator for the independent variable that maximizes the objective function. For the absolute value operator, This is the discrete-time offset index in the cross-correlation sequence, with a value ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. For sensing nodes In the serial number The discrete cross-correlation sequence of the data frames, For the integer identifier of the sensor node, It is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer.

[0081] Indoor sound field propagation exhibits causality and reverberation attenuation characteristics, meaning that sound energy rises rapidly upon reaching the microphone and then attenuates exponentially due to multiple reflections at the room boundaries. Based on this physical characteristic, the sound masking window constructed by the microcontroller has a shorter pre-retention length to the left of the main peak position and a longer post-retention length to the right of the main peak position.

[0082] In the specific underlying implementation, due to discrete cross-correlation sequences It is the cyclic cross-correlation result obtained based on the frequency domain fast Fourier transform, and its discrete-time offset index. The underlying memory has a single frame length The masking window exhibits a cyclic characteristic. If the masking window boundary is calculated directly using linear addition and subtraction, an array addressing out-of-bounds error will occur when the main correlation peak is near the beginning or end of the data frame. Therefore, the microcontroller employs a cyclic boundary determination mechanism that combines modulo operations when constructing the acoustic masking window.

[0083] The mathematical expression for the acoustic masking window function is as follows: ; In the formula, For sensor nodes The A dynamic acoustic masking window function constructed from individual data frames. This is a discrete-time offset index, with values ​​ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. This is the discrete-time offset index corresponding to the main correlation peak. The length reserved before the acoustic masking window, expressed in units of the number of sampling points. The post-retention length of the acoustic masking window, expressed in units of the number of sampling points. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. For modulo mathematical operators.

[0084] In the specific implementation of lower-level features, the pre-reserved length is... This is used to preserve the small energy ramp-up phase at the leading edge of the main peak. This ramp-up phase is primarily caused by the physical group delay characteristics of the anti-aliasing low-pass filter within the analog-to-digital converter. The microcontroller pre-sets the preservation length. The fixed setting is 20 to 50 sampling points. Post-retention length. The late-reflection acoustic energy used to fully cover the room is determined by the preset nominal reverberation time of the target building's receiving room and the system sampling rate, and the calculation relationship is the product of the nominal reverberation time and the system sampling rate.

[0085] The microcontroller performs element-wise multiplication of the dynamically generated acoustic masking window with the discrete cross-correlation sequence. Values ​​outside the masking window are set to zero, thus truncating the background noise.

[0086] The formula for calculating sequence truncation is as follows: ; In the formula, The truncated effective room impulse response sequence, This is the original discrete cross-correlation sequence. For acoustic masking window function, This is a discrete-time offset index, with values ​​ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. For the integer identifier of the sensor node, It is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer.

[0087] Through the aforementioned dynamic construction and truncation mechanism, the microcontroller filters out the tail noise that is irrelevant to the measurement in the local hardware without changing the original physical characteristics of the room's acoustic response, thus providing a data benchmark for subsequent acoustic energy integration.

[0088] For optimizations of low-level assembly instructions for array traversal addressing within a microcontroller and memory alignment mechanisms for window function multiplication, those skilled in the art can consult relevant microprocessor software development manuals, as these are well-known technologies in the field and will not be elaborated upon here.

[0089] The specific implementation of the peak signal-to-noise ratio extraction and local primary decision logic in the building sound insulation performance testing method provided by this invention is as follows: In the measurement principle of acoustic systems, the broadband pseudo-random sequence used in the test has strong autocorrelation and weak cross-correlation. When incoherent transient environmental noise occurs inside the receiving room, after cross-correlation mapping with the conjugate spectrum of the locally stored excitation signal, the energy of this noise is uniformly dispersed and distributed across the entire discrete time offset axis, manifesting as an overall increase in background noise energy. This increase in noise floor leads to a significant decrease in the relative height between the correlation peak formed by the transmitted signal in the discrete cross-correlation sequence and the noise floor. Based on this physical mapping characteristic, the microcontroller extracts the peak signal-to-noise ratio of the discrete cross-correlation sequence to quantify the signal quality of the current data frame, thereby executing local primary decision logic to determine whether the current data frame is contaminated by environmental noise.

[0090] The microcontroller calculates the effective signal energy within the effective room impulse response sequence and the background noise energy outside the masking window, respectively. The effective signal energy is obtained by summing the squares of the amplitudes of the truncated effective room impulse response sequence. The background noise energy is obtained by summing the squares of the amplitudes of the remaining parts after subtracting the truncated effective room impulse response sequence from the original discrete cross-correlation sequence.

[0091] The formulas for calculating signal and noise energy are as follows:

[0092] In the formula, For sensing nodes In the serial number The effective signal energy of the data frame For sensing nodes In the serial number Background noise energy of the data frame The truncated effective room impulse response sequence, This is the original discrete cross-correlation sequence. This is a discrete-time offset index, with values ​​ranging from 0 to... integers, This represents the total number of discrete sampling points contained within a single data frame. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. This is the absolute value operator.

[0093] The microcontroller uses the effective signal energy and background noise energy obtained from the above calculations to calculate the peak signal-to-noise ratio (PSNR) of the current data frame. Since the range of acoustic energy variation in architectural acoustics is quite large, the microcontroller uses logarithmic operations to map it to a decibel value. During the underlying hardware calculation execution phase, to prevent microcontroller hardware floating-point aberrations caused by a zero denominator when the ambient noise is low, the microcontroller adds a preset anti-overflow minimal positive bias constant to the denominator of the logarithmic independent variable.

[0094] The formula for calculating peak signal-to-noise ratio is as follows:

[0095] In the formula, For sensing nodes In the serial number The peak signal-to-noise ratio of the data frame, in decibels. For sensing nodes In the serial number The effective signal energy of the data frame For sensing nodes In the serial number Background noise energy of the data frame To prevent overflow, the minimum positive bias constant is typically set to a value of 10. -6 Up to 10 -8 The floating-point numbers between these values ​​are used to ensure that the accuracy of the normal signal-to-noise ratio calculation is not affected. For the integer identifier of the sensor node, This is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer. It is a base-10 logarithmic operator.

[0096] The microcontroller compares the calculated peak signal-to-noise ratio (SNR) with a preset local decision threshold to output a judgment result. In the specific implementation of the lower-level features, the local decision threshold is set based on the statistical average of the system noise floor under interference-free conditions. Before the formal measurement, the system performs pre-sampling during a period of environmental silence to obtain the baseline SNR of the measurement environment. The microcontroller sets the local decision threshold to a specific value 3 to 5 dB lower than this baseline SNR to distinguish between normal sound field fluctuations and incoherent transient noise.

[0097] The mathematical expression for the local primary decision logic is as follows:

[0098] In the formula, For sensing nodes For serial number The local decision boolean value output by the data frame. For sensing nodes In the serial number The peak signal-to-noise ratio of the data frame. The preset local decision threshold, For the integer identifier of the sensor node, It is the common globally increasing sequence number of the data frame and its corresponding discrete time window, and its value is a non-negative integer.

[0099] When the microcontroller outputs the local decision boolean value When the value is 1, it indicates that the current data frame has not been contaminated by burst noise, and the microcontroller maintains the validity status flag bit of the data frame in the first-in-first-out circular buffer of memory at its initial valid state. When the microcontroller outputs a local decision Boolean value... When the value is 0, it indicates that the current data frame is contaminated by sudden noise. In this case, the microcontroller performs a local data interception operation, overwriting the validity status flag bit corresponding to the data frame in memory with a value of 0, representing an invalid state. The microcontroller triggers the node communication module to broadcast a global veto beacon containing the current globally incrementing sequence number via the radio frequency communication channel, sending a contamination warning to other sensor nodes in the distributed receiving node array.

[0100] For the underlying Taylor series expansion approximation algorithm of the logarithmic function inside the microcontroller and the register configuration method for triggering interrupt signals of the node communication module, those skilled in the art can refer to the relevant embedded processor programming manuals, which are well-known technologies in the field and will not be elaborated here.

[0101] The specific implementation of the deterministic backoff and channel activity detection based on the spatial collaborative veto strategy using node identifier mapping in the building sound insulation performance testing method provided by this invention is as follows: In distributed building acoustics measurements, when incoherent transient environmental noise occurs inside the receiving room, this noise, due to its spatial diffusion characteristics, is typically detected simultaneously by multiple sensor nodes in the distributed receiving node array. If multiple sensor nodes with contaminated decision data frames simultaneously broadcast global veto beacons into the radio frequency channel, air collisions of the radio frequency signals will occur, preventing the beacons from being properly demodulated. Traditional random backoff mechanisms, due to the unpredictability of their backoff times, cannot provide a definite maximum system communication delay, thus risking that the communication delay exceeds the underlying time constraints of the aforementioned circular buffer depth. To address this technical problem, the microcontroller employs a deterministic backoff mechanism based on the mapping of the underlying physical identifiers of the sensor nodes.

[0102] In terms of general mechanism principles, the system directly maps the integer identifiers of sensor nodes to multiples of the backoff time slots. This allows sensor nodes with smaller identifiers to have priority in transmitting on the time axis, while sensor nodes with larger identifiers have sufficient time windows to listen to the channel during the longer backoff waiting period. This identifier-based staggered order mechanism avoids concurrent collisions while providing a mathematically calculable maximum waiting time boundary.

[0103] During the initialization phase, the system assigns a globally unique integer identifier to each sensor node in the distributed receiver node array. After the microcontroller executes its local primary decision logic and outputs the contamination determination result, it does not immediately trigger the node communication module to transmit radio frequency. Instead, it calculates a deterministic backoff delay based on its own integer identifier.

[0104] The formula for calculating deterministic backoff time is as follows: ;

[0105] In the formula, For sensing nodes Deterministic backoff time before broadcasting; An integer identifier for the sensor node, with a value ranging from 1 to... positive integers, This represents the total number of sensor nodes included in the distributed receiver node array. Based on the backoff time slot length.

[0106] In the specific implementation of lower-level features, the basic backoff time slot length The setting is based on the physical hardware time required for the node communication module to complete a full preamble transmission and RF transceiver state switch. To ensure that the transmission windows of different sensor nodes are staggered on the time axis and do not overlap, the microcontroller sets the basic backoff time slot length. The time is typically set between 5 and 10 milliseconds. The maximum deterministic backoff time of the system is... This parameter directly supports the allocation of the depth of the ring buffer slot at the system's underlying level.

[0107] During the period of certainty retreat Subsequently, the microcontroller performs channel activity detection on the current wireless channel via the node communication module. The purpose of channel activity detection is to avoid redundant beacon broadcast operations. Since any global veto beacon emitted by any sensor node contains a globally incrementing sequence number of the current data frame, the distributed receiver array only needs to receive a veto beacon for that sequence number once to discard the contaminated data frame globally.

[0108] To reduce the time overhead of RF demodulation, the microcontroller directly infers whether a sensor node with a smaller integer identifier is already performing a broadcast task by detecting the presence of concurrent RF energy in the channel. Since all sensor nodes in the distributed receiver array maintain strict discrete-time window synchronization based on a global clock, the decision-making and backoff actions for data frames with the same globally incrementing sequence number are strictly constrained within the same physical time span of hundreds of milliseconds. Therefore, bursts of RF energy detected within this preset time window and preset frequency band are highly likely to originate from global rejection beacons emitted by other sensor nodes in the same group. Based on this highly correlated and strictly synchronized physical premise, the microcontroller can directly rely on the underlying channel energy amplitude to make anti-redundancy decisions.

[0109] The microcontroller obtains the energy amplitude of the current radio frequency channel by reading the received signal strength indicator register inside the node communication module, and compares it with the preset channel busy / idle threshold.

[0110] The decision logic formula for the channel state is as follows: ; In the formula, For sensing nodes The current channel busy / idle status is determined by a Boolean value, where 1 indicates the channel is busy and 0 indicates the channel is idle. This is the received signal strength indication value currently measured and acquired by the node communication module, in decibels and milliwatts (dBm). The preset channel busy / idle threshold is expressed in decibels and milliwatts (dBm). An integer identifier for the sensor node, with a value ranging from 1 to... positive integers, where This represents the total number of sensor nodes.

[0111] Channel busy / idle threshold The noise level is determined by the electromagnetic background noise at the measurement site. Before the formal acoustic measurement, the node communication module scans the electromagnetic energy of the current frequency band, and the microcontroller sets the channel busy / idle threshold. The value is set to be 5 to 10 dB higher than the electromagnetic background noise level, with a typical range of -100 dBm to -90 dBm.

[0112] When the channel busy / idle status Boolean value When the value is 1, it indicates that other nodes in the channel are already transmitting a veto beacon. In this case, the microcontroller directly cancels the current broadcast task for this data frame, and the node communication module maintains the receive listening mode; when the channel busy / idle status boolean value is 1, it indicates that other nodes in the channel are already transmitting a veto beacon. When the value is 0, it indicates that the channel is idle. The microcontroller controls the node communication module to immediately modulate and transmit a global veto beacon containing a globally incrementing sequence number to the radio frequency air channel. By combining the above physical identifier mapping with channel energy detection, the system reduces the broadcast power consumption of the node communication module while avoiding co-channel interference collisions.

[0113] For the analog-to-digital conversion sampling process of the internal RF transceiver front-end of the node communication module and the hardware register reading mechanism for the received signal strength indication, those skilled in the art can consult the relevant RF communication chip datasheets, which are well-known technologies in the field and will not be elaborated here.

[0114] The specific implementation of the frame structure design of the global veto beacon based on the spatial collaborative veto strategy of node identifier mapping in the building sound insulation performance testing method provided by this invention is as follows: In a distributed building acoustics measurement system, the global veto beacon is the control command that triggers the spatial collaborative veto mechanism. Since each sensor node collects a continuous, uninterrupted stream of audio data, the physical space capacity of the microcontroller's built-in static random access memory is limited. If the wireless network communication protocol uses a packet structure with a variable-length extension field or requires multi-hop routing, the processing time of the media access control layer and the radio frequency over-the-air transmission become unpredictable. Unpredictable communication latency means that the microcontroller cannot accurately calculate and allocate the safe slot depth of the FIFO circular buffer, leading to memory overflow or contaminated data frames being passed through to subsequent computing units before being vetoed. Therefore, to meet the underlying time constraints of the FIFO circular buffer depth and ensure that the global veto beacon is received and parsed by all remote nodes before the contaminated data frame is removed from the FIFO circular buffer exit, the global veto beacon constructed by the microcontroller adopts a fixed-length simplified structure, without routing information or a variable-length extension field.

[0115] The frame structure of the global veto beacon consists of four consecutive fields: physical layer preamble, network synchronization word, core data payload, and cyclic redundancy check code.

[0116] The physical layer preamble is used to wake up the RF front-end circuitry of the receiver node's communication module and to achieve bit synchronization of the underlying clock. The microcontroller sets the physical layer preamble to an alternating binary sequence with a fixed length of 8 bytes. The network synchronization word is used to distinguish the current acoustic test network from other wireless networks coexisting in the measurement field, preventing cross-network interference. The network synchronization word consists of a fixed network identifier uniformly allocated during system initialization, occupying 2 bytes.

[0117] The core data payload is the effective payload area for transmitting veto commands. To reduce radio frequency transmission time, the core data payload is limited to 4 bytes. The first two bytes carry the integer identifier of the sensor node initiating the broadcast, for the receiving sensor node to record the source and perform system diagnostics. The last two bytes carry the globally incrementing sequence number of the data frame locally determined to be contaminated. This globally incrementing sequence number is the unique identifier for the distributed receiving node array to collaboratively match the target data frame within the spatial range.

[0118] Cyclic Redundancy Check (CRC) codes are appended to the core data payload, occupying 2 bytes. The microcontroller calls its internal hardware check unit to perform a polynomial division operation on the network synchronization word and the core data payload to generate a check sequence. The receiver compares this check sequence to identify and directly discard erroneous frames caused by channel fading during radio frequency transmission.

[0119] Through the aforementioned lightweight structural design, the total number of bytes for the global veto beacon is fixed as a constant. The microcontroller, in conjunction with the modulation and demodulation rate of the underlying physical layer of the node communication module, calculates the radio frequency air transmission time of the global veto beacon.

[0120] The formula for calculating the time taken for radio frequency air transmission is as follows: ; In the formula, The time taken for the radio frequency air transmission of the global denial beacon, in seconds. This is the length of the physical layer preamble, fixed at 8 bytes. The length of the network synchronization word is fixed at 2 bytes. The length of the core data payload is fixed at 4 bytes. The length of the cyclic redundancy check (CRC) code is fixed at 2 bytes. This is a fixed number of bits contained in a single byte, with a constant value of 8. The radio frequency transmission bit rate configured for the node communication module at the physical layer is typically set to a range of 50,000 to 250,000 bits per second (bps). The specific value of this parameter is determined by the physical communication distance and the radio frequency band at the measurement site.

[0121] Radio frequency over-the-air transmission time of global veto beacon Together with the aforementioned maximum deterministic backoff time, this parameter constitutes the maximum total communication delay of the system, which is the technical basis for the allocation of the underlying circular buffer in the system.

[0122] The formula for calculating the maximum total communication delay of the system is as follows: ; In the formula, The maximum total communication delay of the system is expressed in seconds. The total number of sensor nodes in the distributed receiver node array, taking a positive integer value. The basic backoff time slot length, in seconds. The time taken for radio frequency air transmission of the global veto beacon, in seconds.

[0123] The microcontroller is based on the system's maximum total communication latency. During system initialization, the physical slot depth of the first-in-first-out (FIFO) circular buffer in the statically allocated memory is determined. The underlying constraints must ensure that the total physical time span of the circular buffer is strictly greater than the system's maximum total communication latency. This ensures sufficient time for node-to-node collaborative communication through physical mechanisms.

[0124] After assembling the underlying frame of the global veto beacon, the microcontroller writes it into the transmit buffer register of the node communication module and triggers an RF transmission command. The node communication module modulates the global veto beacon containing a pollution warning into a high-frequency electromagnetic wave and transmits it to the space channel. Other sensor nodes distributed at the measurement site, after receiving the global veto beacon and completing verification, extract the globally incrementing sequence number and then perform a coordinated interception operation on the data frame corresponding to that sequence number in their own first-in-first-out circular buffer.

[0125] For the polynomial generator matrix of the cyclic redundancy check code inside the microcontroller and the physical implementation of frequency shift keying modulation and demodulation at the bottom layer of the node communication module, those skilled in the art can refer to relevant digital communication protocols and integrated circuit application notes, which are well-known technologies in the field and will not be elaborated here.

[0126] The specific implementation method of pointer tracing and forced overwriting of the cached state of remote nodes based on the spatial collaborative veto strategy of node identifier mapping in the building sound insulation performance testing method provided by the present invention is as follows: After the node communication module of the receiving sensor node detects the global veto beacon in the radio frequency air channel, it performs physical layer demodulation and cyclic redundancy check (CRC). Upon successful CRC check, the node communication module triggers an external hardware interrupt on the microcontroller. The microcontroller responds to the interrupt request, extracts the core data payload of the global veto beacon from the node communication module's receive buffer register, and parses out the globally incrementing sequence number of the data frame determined to be contaminated by the remote node. This sequence number will serve as the traceability key for the target data frame in local memory.

[0127] At the underlying data structure management level, data frames in the first-in-first-out (FIFO) circular buffer are physically stored sequentially according to their chronological order. Because the sensor nodes in the distributed receiver array maintain strict discrete-time window synchronization, there is a strict linear mapping relationship between the globally incrementing sequence number of the data frame and the physical address of the buffer. Therefore, the microcontroller does not need to traverse the entire memory space to compare and search for the target sequence number. Instead, it uses the difference in sequence numbers as a relative offset and combines it with modulo operations to directly calculate the physical memory address of the target data frame. This pointer-direct addressing mechanism, with constant time complexity, reduces the execution time of the microcontroller's interrupt service routine and avoids the risk of frame loss during ongoing audio sampling due to excessively long table lookup times.

[0128] All sensor nodes in the distributed receiver array maintain synchronized sampling within discrete time windows based on a global clock. Data frames acquired and generated by each sensor node at the same time are assigned the same globally incrementing sequence number. The microcontroller utilizes this global synchronization characteristic to calculate the offset between the sequence number of the received target data frame and the sequence number of its most recently processed local data frame.

[0129] The formula for calculating the serial number offset is as follows: ; In the formula, This is the serial number offset, and its value is an integer. This is the globally incrementing sequence number of the latest generated data frame at the local sensor node, and its value is a non-negative integer. This is the globally incrementing sequence number of the contaminated data frame extracted from the global veto beacon, and its value is a non-negative integer.

[0130] The microcontroller performs pointer tracing within a FIFO circular buffer in memory based on the calculated sequence number offset. This FIFO circular buffer represents a contiguous address space in physical memory, which the microcontroller manages by maintaining a continuously moving write pointer. Using the current write pointer index, the sequence number offset, and the physical slot depth of the circular buffer, the microcontroller calculates the absolute storage index of the target data frame in memory.

[0131] The formula for calculating the target storage index is as follows: ; In the formula, This is the storage index of the target data frame in the first-in-first-out circular buffer, and its value ranges from 0 to... integers, This is the write pointer index corresponding to the latest written data frame in the first-in-first-out circular buffer, and its value ranges from 0 to... integers, The obtained sequence number offset is calculated and takes an integer value. The physical slot depth of the first-in-first-out circular buffer is a positive integer, and its specific value is determined by rounding up the ratio of the maximum total system communication delay to the length of a single frame time window. For modulo mathematical operators.

[0132] After obtaining the target storage index, the microcontroller must perform a boundary security check on the validity of the relative offset. If the sequence number offset... Greater than or equal to the physical slot depth This indicates that the target data frame, due to its age, has been physically overwritten and removed from memory by newly acquired data frames; if A value less than 0 indicates network out-of-order behavior or a local clock synchronization anomaly, resulting in the reception of a rejection beacon for a future time window. In either of these out-of-bounds scenarios, the microcontroller directly abandons the overwrite operation and exits the interrupt service routine. If the sequence number offset... Less than the physical slot depth If the value is greater than or equal to 0, it indicates that the target data frame is still residing in the local first-in-first-out circular buffer and awaiting further processing.

[0133] Under conditions that boundary safety is satisfied, the microcontroller performs a forced overwrite operation. The microcontroller determines this based on the target memory index. The physical address for direct addressing is generated by combining the base address of the memory. The structure member of the target data frame at that address is accessed, and its validity status flag is forcibly overwritten from the initial valid state to the invalid state.

[0134] The underlying logic expression for forced overriding is as follows: ; In the formula, For sensing nodes The index in the local first-in-first-out circular buffer is The validity status flag bit of the data frame indicates that a value of 0 represents an invalid status and a value of 1 represents a valid status. This is the storage index of the target data frame, and its value ranges from 0 to... integers, This is an integer identifier for the local sensor node, and its value is a positive integer.

[0135] Through the aforementioned pointer tracing and forced overwrite mechanism, any local burst noise detected by any sensor node in the distributed receiving node array can be converted into a synchronous interception operation targeting data frames of the same time period across the entire network. When a data frame with an invalid status flag is normally moved out of the FIFO circular buffer over time and is about to enter the subsequent acoustic energy integration calculation unit, the microcontroller will first read the status flag. Once a value of 0 is detected, the microcontroller will execute a conditional jump, skipping the acoustic energy integration accumulation calculation of that frame, thereby completely eliminating data segments contaminated by transient environmental noise in the final sound insulation and transmission acoustic energy measurement results. This processing mechanism achieves spatially distributed cooperative anti-interference using only short-byte beacons without transmitting high-frequency raw audio data.

[0136] For the register mapping of the microcontroller's internal and external hardware interrupt vector tables and the low-level assembly operations of direct addressing of memory structure pointers, those skilled in the art can consult relevant microprocessor architecture design guidelines, which are well-known technologies in the field and will not be elaborated here.

[0137] The specific implementation methods for verifying the validity of asynchronous pop-out integrals and multi-point spatial averaging solutions, and for accumulating equivalent acoustic energy integrals in the building sound insulation performance testing method provided by this invention are as follows: In a distributed building acoustics measurement system, the FIFO circular buffer within the memory decouples data latency from physical time. When a data frame's residence time in the buffer reaches the maximum total communication latency determined by the physical slot depth, the data frame is popped from the underlying physical memory as the read pointer advances. At this point, through a joint screening process involving local primary decision logic and a remote spatial collaborative veto mechanism, the final determination of whether the data frame is contaminated by transient environmental noise has been made.

[0138] In general architectural acoustics measurement principles, evaluating the sound insulation performance of a wall depends on the average acoustic energy of the received steady-state sound field within the room. Because anti-interference mechanisms dynamically and discretely discard contaminated data frames along the time axis, the total duration of effective audio data actually retained and integrated by each sensor node distributed in space at the end of the measurement cycle is inconsistent. If the accumulated total acoustic energy is directly spatially calculated, the unequal integration time will directly translate into a calculation deviation in the acoustic energy distribution. To eliminate the error caused by this time length difference, the microcontroller employs an asynchronous pop-out conditional integration accumulation mechanism. By independently recording the accumulated acoustic energy value and the number of effective data frames, it uses division operations to equivalently reconstruct the steady-state average acoustic energy of a single frame, thereby mathematically repairing the time axis break caused by discarded data frames.

[0139] During the memory data reading phase, the microcontroller maintains a read pointer for a first-in-first-out (FIFO) circular buffer. When the system global clock triggers a read cycle, the microcontroller accesses the corresponding memory address based on the read pointer index and retrieves the data about to be popped from the stack. The validity status flag of each data frame is determined and loaded into the condition judgment register.

[0140] During the stack pop validity check, the microcontroller performs a conditional branch check based on the validity status flag. If the read validity status flag is 0, it indicates that the data frame has been determined to be contaminated by sudden environmental noise during the previous waiting period, either locally or remotely. In this case, the microcontroller directly discards the data frame, releases the corresponding memory slot, and skips the sound energy integration and accumulation operation. If the read validity status flag is 1, it indicates that the data frame has not been affected by any sudden noise interference during the entire determination period. The microcontroller sends the valid data frame and its associated valid signal energy to the accumulation calculation unit.

[0141] After completing the validity check, the microcontroller maintains two global accumulators in parallel throughout the entire acoustic measurement cycle using cyclic accumulation instructions. One global accumulator is used to calculate the sum of valid signal energy, and the other global accumulator is used to count the total number of valid data frames actually involved in the integration.

[0142] The formulas for calculating the effective signal energy accumulation and the effective frame count are as follows: ; ; In the formula, For sensing nodes The total effective transmitted sound energy accumulated over the entire measurement period. This represents the total number of discrete data frames generated by the sensor node within the current measurement cycle. It is a positive integer, and its specific value is determined by dividing the total duration of the entire acoustic measurement cycle by the physical duration of a single discrete time window. For sensing nodes The The validity status flag bit is used when a data frame is popped from the stack. A value of 0 indicates an invalid status, and a value of 1 indicates a valid status. For sensing nodes The The effective signal energy of each data frame For sensing nodes The total number of valid data frames actually involved in acoustic energy integration during the measurement period, taking a non-negative integer value. This is an integer identifier for the sensor node, and its value is a positive integer. is the globally increasing sequence number of the discrete time window, and its value is a positive integer.

[0143] After the measurement cycle ends or the test signal stops being transmitted, the microcontroller calculates the equivalent time-averaged acoustic energy at the spatial location corresponding to the sensing node based on the final values ​​of the two global accumulators mentioned above. To prevent all data frames from being judged as contaminated and resulting in a total number of valid data frames of zero under extremely harsh environments, the microcontroller introduces a very small positive bias constant in the denominator of the division operation to prevent memory overflow during division by zero.

[0144] The formula for calculating the equivalent time-averaged sound energy is as follows: ; In the formula, For sensing nodes The equivalent time-averaged acoustic energy was calculated. For sensing nodes The total effective transmitted sound energy accumulated over the entire measurement period. For sensing nodes The total number of valid data frames actually involved in acoustic energy integration during the measurement period, taking a non-negative integer value. To prevent overflow, the minimum positive bias constant is typically set to a value of 10. -6 Up to 10 -8 floating-point numbers between This is an integer identifier for the sensor node, and its value is a positive integer.

[0145] Using this asynchronous integration algorithm with conditional masking features, the microcontroller effectively removes the distortion energy caused by sudden disturbances from the discrete time series. This equivalent time-averaged sound energy accurately reflects the sound insulation and transmission characteristics of the target test wall under steady-state excitation conditions, laying a data foundation for subsequent spatial sound field averaging and sound pressure level mapping.

[0146] For the register overflow handling mechanism of the floating-point accumulation calculation unit inside the microcontroller and the single-precision and double-precision floating-point number type conversion operation, those skilled in the art can refer to the relevant digital signal processing implementation guides and microprocessor development manuals. These are well-known technologies in the field and will not be elaborated here.

[0147] The specific implementation methods for asynchronous compliance determination and network sleep control in the building sound insulation performance testing method provided by this invention are as follows: In architectural acoustics measurement, to ensure the statistical stability of the transmitted sound field of the target test wall over time, the system needs to acquire effective acoustic signals that meet a minimum time length. Because a spatial cooperative veto mechanism dynamically and discretely removes contaminated data frames along the time axis, the number of effective data frames actually retained and integrated by each sensor node in the distributed receiver array within the same physical time exhibits spatial distribution differences. This data retention difference causes the time required for sensor nodes in different spatial locations to complete the minimum effective acoustic energy integration to be asynchronous.

[0148] In typical multi-point measurement systems, if a globally fixed-duration synchronous measurement mechanism is used, sensor nodes in low-interference areas, after completing their data accumulation tasks in advance, still need to passively maintain acoustic sampling and RF channel listening. This not only increases the node's own power consumption but also continuously occupies the wireless network's media access time. Since the equivalent time-averaged acoustic energy of each spatial measurement point is calculated and statistically analyzed independently, a measurement point reaching the effective time threshold means that the acoustic characteristics at that location have converged and meet the standard requirements. Therefore, the microcontroller is configured with an asynchronous compliance determination mechanism, allowing each node to independently exit the network. This mechanism not only reduces power consumption but also proactively releases RF channel resources for other sensor nodes that are still performing measurement tasks due to significant interference, reducing the probability of subsequent air collisions.

[0149] At the underlying data management mechanism, the microcontroller independently monitors the total number of valid data frames actually participating in acoustic energy integration at local sensor nodes. During the initialization phase, the system statically allocates a target valid frame count threshold. The specific value of this threshold is determined by dividing the minimum valid integration time required by the acoustic measurement standard by the physical duration of a single discrete time window and rounding down. In conventional architectural acoustic measurement specifications, the minimum valid integration time is set to a value between 6 and 15 seconds. After each valid data frame is popped from the stack and the acoustic energy integration is completed, the microcontroller uses a hardware comparator to perform a compliance determination.

[0150] The logic formula for determining whether the standard has been met is as follows: ; In the formula, For sensing nodes The local points threshold status is determined by a Boolean value, where 1 indicates that the threshold has been met and 0 indicates that the threshold has not been met. For sensing nodes The total number of valid data frames actually involved in acoustic energy integration during the measurement period, taking a non-negative integer value. The preset target effective frame count threshold, with a value that is a positive integer. This is an integer identifier for the sensor node, with a value of positive integer ranging from 1 to the total number of sensor nodes.

[0151] When the local integration threshold boolean value flips from 0 to 1, it indicates that the sensor node has accumulated sufficient duration of transmitted audio data free from environmental noise pollution. At this point, the microcontroller blocks the data acquisition interruption of the front-end analog channel and stops the generation of new data frames within the discrete time window. The microcontroller extracts the total accumulated effective transmitted sound energy and the total number of effective data frames stored in the global accumulator, and executes a division instruction to calculate the equivalent time-averaged sound energy. Subsequently, the microcontroller encapsulates the equivalent time-averaged sound energy and the integer identifier of the local sensor node into a final measurement result data packet, and controls the node communication module to send the result data packet to the system's main control aggregation device via the radio frequency air channel.

[0152] After the confirmation result data packet is successfully received by the master control device and a media access control layer confirmation frame is returned, the microcontroller executes the underlying network sleep control to cut off the power supply circuit of non-essential peripherals.

[0153] The power state transition logic formula for the sensor node is as follows: ; In the formula, For sensing nodes The current system power supply status is as follows: This represents the normal operating power consumption state when global sampling and radio frequency communication are in active mode. To disable the deep sleep low-power state of the analog-to-digital converter, microphone bias circuit, and node communication module RF receiver front end, For sensing nodes The local points threshold status is determined by a Boolean value, where 1 indicates that the threshold has been met and 0 indicates that the threshold has not been met. This is an integer identifier for the sensor node, with a value of positive integer ranging from 1 to the total number of sensor nodes.

[0154] By implementing the aforementioned asynchronous compliance determination and network sleep control strategy, each sensor node within the distributed receiving node array independently exits the sound field measurement network after completing its own effective acoustic energy accumulation task. This mechanism avoids redundant measurement waiting power consumption at the physical level, and provides hardware-level support for the power supply endurance of distributed wireless sensor nodes during long-term building sound insulation testing tasks.

[0155] For the clock gating configuration of the power management unit inside the microcontroller, the bit operations of the peripheral power status register, and the sleep / wake instruction sequence of the RF communication chip, those skilled in the art can consult the relevant integrated circuit datasheets and low-power system architecture design guidelines. These are well-known technologies in the field and will not be elaborated here.

[0156] The specific implementation method of low-bandwidth data backhaul and spatial energy calculation for asynchronous pop-out integration and multi-point spatial averaging in the building sound insulation performance testing method provided by this invention is as follows: In general architectural acoustic measurements, due to standing waves and sound field interference within the three-dimensional space of a room, the sound pressure level measured at a single point cannot represent the transmitted sound energy distribution of the entire receiving room. To obtain accurate sound insulation, the system must perform spatial equivalent averaging of acoustic measurement data from multiple sensor nodes scattered at different locations within the receiving room. In traditional centralized processing architectures, all nodes need to transmit the high-frequency sampled raw audio data stream concurrently to the main control unit in real time, which can easily lead to severe congestion of the wireless channel and cause RF signal attenuation and packet loss when penetrating soundproof walls with high surface density. The underlying technical logic of this invention performs the aforementioned local cooperative anti-interference and asynchronous integration operations in the microcontrollers of each sensor node, converging a long-duration discrete audio sequence into a single sound energy scalar. This edge computing mechanism eliminates the need for wireless transmission of the raw audio data, fundamentally enabling low-bandwidth data backhaul that supports through-wall communication.

[0157] In the data return phase, after calculating the equivalent time-averaged acoustic energy, the microcontroller of the sensor node encapsulates the floating-point acoustic energy data with its own integer identifier into a fixed-length result data packet. Due to the small payload of the result data packet, the microcontroller configures the node communication module to use a low-bit-rate physical layer modulation mode for transmission. This reduced RF transmission rate proportionally improves the demodulation sensitivity and multipath fading resistance of the node communication module's receiving front end, enabling the data packet to stably penetrate the building's soundproof walls using the limited link budget and be successfully received by the system's main control aggregation device.

[0158] In terms of spatial averaging principles, since the sound field at multiple points within the receiving room is a complex, random reverberation field formed by multiple reflections of transmitted sound waves, the sound pressure signals collected by each sensor node are uncorrelated in phase. Therefore, at the physical level, a simple arithmetic average of the sound pressure amplitude or logarithmic decibel value cannot be directly performed. Instead, it is necessary to first preserve it as an energy physical quantity proportional to the sound energy at each measurement point, and then perform a linear arithmetic average calculation in the spatial dimension. Within a preset measurement period, the main control aggregation device listens to and collects the result data packets transmitted from all sensor nodes in the distributed receiving node array. To ensure the data integrity of the spatial calculation, the main control aggregation device maintains a topology registry containing the identifiers of all sensor nodes participating in the measurement. After comparing the topology registry and confirming that the equivalent time-averaged sound energy of all measurement points has been collected, the processor inside the main control aggregation device performs multi-point spatial energy calculation to obtain the spatial average sound energy within the receiving room.

[0159] The formula for calculating the average acoustic energy in space is as follows: ; In the formula, The average acoustic energy of the receiving room space is calculated and obtained by the main control aggregation equipment. The total number of sensor nodes in the distributed receiver node array, taking a positive integer value. For the identifier is The equivalent time-averaged acoustic energy transmitted back by the sensor node. This is an integer identifier for the sensor node, with a value of a positive integer ranging from 1 to... .

[0160] After acquiring the spatial average acoustic energy, the main control convergence device needs to map it to a spatial average sound pressure level that can be directly read in engineering, according to the definition in the building acoustics code. This logarithmic conversion process depends on the acoustic calibration constants at the system's underlying level.

[0161] The formula for calculating the space-averaged sound pressure level is as follows: ; In the formula, The final calculated output is the average sound pressure level in the receiving room, expressed in decibels (dB). For acoustic decibel conversion, a fixed constant. For base-10 logarithmic operations, To calculate the average acoustic energy of the receiving room space, The system's preset reference acoustic energy constant is a physical quantity that is equivalent to the electrical signal energy generated by a reference sound pressure of 20 μPa in a standard air medium.

[0162] Reference acoustic energy constant The specific values ​​are determined during the hardware calibration phase before the formal acoustic measurements. Calibrators use a standard sound level calibrator to apply a constant pure-tone excitation with a known sound pressure level (typically 94 dB at a frequency of 1 kHz) to the microphone front-ends of the sensor nodes in the distributed receiver node array. The main control aggregation device collects the digital energy response under this calibration excitation and derives the reference acoustic energy constant by substituting it back into the aforementioned logarithmic formula. The data is stored in non-volatile memory for later use during measurement and calculation. Through this spatial energy calculation process, the system integrates effective acoustic energy data from multiple spatial dimensions and outputs sound pressure level parameters for evaluating the sound insulation performance of walls.

[0163] For the acknowledgment and retransmission mechanism based on the media access control layer at the bottom layer of the master control aggregation device, as well as the hardware coprocessing implementation of the floating-point logarithmic arithmetic instructions inside the microprocessor, those skilled in the art can refer to the relevant wireless network protocol specifications and microcontroller application guidelines. These are well-known technologies in the field and will not be elaborated here.

[0164] See attached document Figure 3 , Figure 3 This is an experimental verification and effect comparison diagram of a building sound insulation measurement system according to an embodiment of the present invention under sudden noise interference, wherein... Figure 3 (a) is a comparison curve showing the change in the measurement error of the average sound pressure level in the receiving room with the frequency of sudden noise. Figure 3 (b) A comparison curve showing the cumulative data transmission volume of the system's wireless network over time. The specific implementation method for experimental verification and effect comparison of the building sound insulation performance testing method provided by this invention is as follows: To verify the effectiveness of this invention, a distributed building sound insulation performance testing system was constructed in a reverberation chamber of an acoustic laboratory conforming to the international standard ISO 10140. The experimental environment included a sound-generating chamber and a receiving chamber, with a single-layer homogeneous glass test wall of known standard sound insulation installed at the test opening between the two chambers. In the receiving chamber, a distributed receiving node array containing six sensing nodes was irregularly arranged, and a main control aggregation device was deployed. The experiment artificially introduced transient acoustic pulses, such as simulated door closing sounds and falling heavy objects, into the receiving chamber to construct a harsh measurement condition with sudden environmental noise interference.

[0165] In the comparison scheme setup, three measurement modes were set up: a baseline interference-free mode, a traditional centralized measurement mode, and the distributed collaborative measurement mode proposed in this invention. The baseline interference-free mode performs measurements against an absolutely quiet background, and its output spatial average sound pressure level serves as the true baseline reference value. The traditional centralized measurement mode lacks inter-node collaborative anti-interference capabilities; each sensor node directly converts high-frequency audio sampling data into sound energy and performs a globally fixed-duration hard integration. The distributed collaborative measurement mode of this invention fully utilizes a microcontroller-based first-in-first-out circular buffer spatial collaborative veto strategy, asynchronous stack-out integration, and network sleep control.

[0166] In the acoustic measurement accuracy verification stage, the system recorded the average sound pressure level in the receiving room at different transient noise occurrence frequencies and calculated the measurement error between it and the benchmark reference value. (See attached...) Figure 3 As shown in (a), the traditional centralized measurement mode cannot identify and remove contaminated audio data segments, and the distortion energy of transient noise is directly mixed into the long-time integrator. As the frequency of transient noise increases, the spatial average sound pressure level calculated by the traditional mode is significantly higher than the reference true value, and its measurement error curve shows a clear upward trend. At the test point where 20 transient noises occur per minute, the maximum positive measurement error reaches 4.5 dB, which seriously masks the true transmitted sound energy characteristics of the test wall, resulting in a lower calculated sound insulation.

[0167] In contrast, the distributed cooperative measurement mode of this invention strictly implements pointer tracing and forced overwriting of the validity status flag bit at the microcontroller level. Interference-affected discrete data frames are precisely bypassed and eliminated by conditional branch instructions before entering the floating-point accumulation unit. Simultaneously, the microcontroller independently counts the total number of valid data frames and performs division operations to equivalently reconstruct the steady-state average sound energy. Experimental results show that even under extreme conditions with 20 high-intensity transient interferences per minute, the error between the spatial average sound pressure level output by this invention and the reference value is consistently suppressed to within 0.25 dB. This low fluctuation error fully demonstrates that the asynchronous pop-out conditional integration algorithm of this invention mathematically repairs the time axis break caused by data frame discarding, ensuring the physical accuracy of multi-point spatial averaging calculations.

[0168] In the system network communication overhead and power consumption assessment phase, the cumulative data transmission volume of the radio frequency air channel was continuously monitored and statistically analyzed throughout the entire measurement period. (See attached...) Figure 3 As shown in (b), if the traditional architecture for transmitting raw audio data is adopted, when the six sensor nodes are working concurrently, the cumulative data transmission volume of the network exhibits a steep linear growth trend due to the limitations of the high-frequency sampling rate and quantization bit depth. The final cumulative transmission volume reaches approximately 15.6 megabytes (MB) within a 60-second measurement period. This high-intensity data concurrency can easily exceed the wireless link capacity limit in the complex multipath environment inside a building.

[0169] This invention converges massive temporal audio sequences into a single equivalent time-averaged acoustic energy scalar at the edge computing side of each sensing node. Throughout the entire measurement period, the sensing node only broadcasts a short global veto beacon to the air channel and sends a fixed-length result data packet once the integration threshold is met. (See attached diagram) Figure 3 As shown in the curve characteristics in (b), the cumulative data transmission volume of this invention is relatively flat in the early and middle stages of measurement, containing only minor beacon overhead. When the measurement reaches about 45 seconds, some sensor nodes in low-interference areas first meet the minimum effective integration time and asynchronously reach the target, triggering a concentrated return of the result data packets, resulting in a slight step increase in the transmission volume curve. Even so, at the end of the 60-second measurement period, the cumulative data transmission volume of this invention remains at the level of several kilobytes (KB), which is more than two orders of magnitude lower than that of the traditional architecture. The lower network bandwidth usage allows the node communication module to use low-rate modulation technology with a high spreading factor, significantly improving the link budget for through-wall communication.

[0170] Furthermore, under the control of the asynchronous compliance determination mechanism, the sensing nodes in the interference-free area autonomously disconnect the power supply circuits between the analog-to-digital converter and the RF receiving front-end after accumulating the minimum effective integration time specified by the standard, entering a deep sleep state. Experiments showed that the overall effective working time of the distributed receiving node array was shortened, the average power consumption of the nodes was reduced, and the battery life of the portable building acoustics detection system was effectively extended. The above experimental results verify the implementation effect of the technical solution proposed in this invention from three dimensions: acoustic accuracy, communication stability, and hardware energy efficiency.

Claims

1. A method for testing the sound insulation performance of a building, characterized in that, Includes the following steps: The master control unit broadcasts precise time protocol frames and plays broadband pseudo-random sequence excitation signals in the sound source room of the target building; Multiple sensor nodes in the receiving room of the target building receive the precise time protocol frame to calibrate the local real-time clock and establish a discrete time window sequence. The collected sound pressure signal is divided into data frames, and after being assigned a sequence number and validity status flag, it is pushed into the local first-in-first-out circular buffer. The sensing node performs cross-correlation operation on the data frame and the copy of the broadband pseudo-random sequence excitation signal stored in the local non-volatile memory to calculate the peak signal-to-noise ratio. If it is lower than the minimum signal-to-noise ratio threshold for determining a valid signal, the validity status flag bit is overwritten to invalid status. After calculating and experiencing a deterministic backoff time, a global veto beacon containing the sequence number is broadcast when the radio frequency channel is idle. After receiving the global veto beacon, the remaining sensor nodes will forcibly overwrite the validity status flag bit of the sequence number in their local first-in-first-out circular buffer to an invalid status. The sensing node performs logarithmic integral calculation of acoustic energy on the data frame that maintains an effective state, and transmits equivalent continuous sound pressure level data back to the main control terminal when the test duration lower limit issued by the system control terminal is reached. The main control unit combines the equivalent continuous sound pressure level data with the sound pressure level data of the sound source room synchronously collected by the reference microphone in the sound source room to calculate and obtain the sound insulation performance evaluation index.

2. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The steps for the sensing node to receive the precise time protocol frame to calibrate its local real-time clock and establish a discrete time window sequence specifically include: The microcontroller of the sensing node records the local timestamp of the arrival time of the RF reception hardware interrupt signal generated by the node communication module inside the sensing node, parses the precise time protocol frame to obtain the transmission time timestamp recorded by the master control end, and calculates the time offset that needs to be compensated between the local real-time clock and the global reference time based on the fixed propagation delay constant calibrated based on the physical distance between the master control end and the sensing node. The microcontroller compensates the time offset into the register of the local real-time clock; After the time offset is compensated, the microcontroller continuously generates internal timing interrupts based on the time window length calculated by dividing the analog-to-digital converter sampling rate by the number of sampling points per frame. It divides the continuous physical time into discrete time windows of equal length and generates the discrete time window sequence by combining the starting absolute physical time parsed from the precise time protocol frame.

3. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The step of the sensing node pushing the data frame into the local first-in-first-out circular buffer specifically includes: The microcontroller of the sensing node allocates contiguous memory blocks in the physical memory space as a local first-in-first-out circular buffer. The storage space allocated to each slot is equal to the product of the total number of discrete sampling points contained in a single data frame and the data bit width of a single sampling point, plus memory space for storing the validity status flag bit. The microcontroller sets the total number of slots in the local first-in-first-out circular buffer to a value corresponding to a physical dwell time greater than the maximum total system communication delay. The physical dwell time is the product of the total number of slots and the time window length. The maximum total system communication delay is the sum of the maximum computation time of the microcontroller, the maximum deterministic backoff time of the sensor node, the radio frequency over-the-air transmission time of the global veto beacon, and the microcontroller processing time of the remaining sensor nodes.

4. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The specific steps of the sensing node performing cross-correlation calculations on the data frame and the broadband pseudo-random sequence excitation signal include: The microcontroller of the sensing node reads the reference conjugate spectrum data stored in the non-volatile flash memory space. The reference conjugate spectrum data is static data generated by performing a discrete Fourier transform on the broadband pseudo-random sequence excitation signal and obtaining the complex conjugate. The microcontroller calls the built-in hardware floating-point unit and uses the fast Fourier transform algorithm to map the data frame from the time domain to the frequency domain to obtain the complex spectrum; The microcontroller performs point-by-point complex multiplication of the complex spectrum with the reference conjugate spectrum data to obtain the cross power spectrum; The microcontroller performs a fast inverse discrete Fourier transform on the cross power spectrum and remaps it back to the time domain to obtain the discrete cross-correlation sequence after dimensionality reduction.

5. The method for testing the sound insulation performance of a building according to claim 4, characterized in that, The step of calculating the peak signal-to-noise ratio specifically includes constructing a noise floor extraction model containing an acoustic masking window, the specific process of which is as follows: The microcontroller traverses the discrete cross-correlation sequence and extracts the discrete time offset index corresponding to the point with the largest absolute value amplitude as the position of the main correlation peak. The microcontroller constructs an asymmetric acoustic masking window based on the position of the correlation peak. The acoustic masking window adopts a cyclic boundary determination mechanism that combines modulo operation. A pre-retention length with a fixed number of sampling points is configured on the left side of the correlation peak position, and a post-retention length determined by the nominal reverberation time of the receiving room of the target building is configured on the right side. The microcontroller multiplies the acoustic masking window with the discrete cross-correlation sequence point by point, sets the values ​​outside the masking window to zero, and obtains the effective room impulse response sequence. The microcontroller calculates the sum of squared amplitudes within the effective room impulse response sequence as the effective signal energy, and the sum of squared amplitudes of the remaining portion after subtracting the effective room impulse response sequence from the original discrete cross-correlation sequence as the background noise energy. After adding a minimum positive bias constant to the effective signal energy and the background noise energy respectively, the microcontroller calculates the ratio of the effective signal energy after adding the minimum positive bias constant to the background noise energy after adding the minimum positive bias constant, and performs a logarithmic operation on the ratio to obtain the peak signal-to-noise ratio.

6. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The step of the sensing node calculating the deterministic backoff time and listening to the radio frequency channel after the deterministic backoff time specifically includes: The microcontroller of the sensing node directly maps the integer identifier of the sensing node to a multiple of the backoff time slot, and obtains the deterministic backoff time by multiplying the integer identifier by the basic backoff time slot length. After the deterministic backoff time, the microcontroller obtains the current energy amplitude of the radio frequency channel by reading the received signal strength indicator register inside the node communication module inside the sensing node; The microcontroller compares the energy amplitude with the channel busy / idle threshold, which characterizes the background noise energy when the channel is not communicating. If the energy amplitude is greater than or equal to the channel busy / idle threshold, the radio frequency channel is determined to be busy and the broadcast operation is canceled. If the energy amplitude is less than the channel busy / idle threshold, the radio frequency channel is determined to be idle and the node communication module is triggered to broadcast the global veto beacon.

7. The method for testing the sound insulation performance of a building according to claim 6, characterized in that, The global veto beacon adopts a fixed-length frame structure without routing information and without variable-length extension fields. The fixed-length frame structure continuously includes: Fixed-byte physical layer preamble; A fixed-byte network synchronization word consists of a fixed network identifier allocated by the master terminal during the network initialization phase; The core data payload is of fixed byte length, wherein the first byte carries the integer identifier of the sensor node that initiated the broadcast, and the last byte carries the sequence number of the data frame that is determined to be contaminated; A fixed-byte cyclic redundancy check (CRC) code is generated by the microcontroller of the sensor node performing a polynomial division operation on the network synchronization word and the core data payload.

8. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The step of the remaining sensing nodes forcibly overwriting the validity status flag bit of the data frame with the sequence number in the local first-in-first-out circular buffer to an invalid status specifically includes: The microcontrollers of the remaining sensor nodes respond to hardware interrupts by extracting the sequence number in the global veto beacon as the target sequence number and calculating the difference between the global incrementing sequence number of the latest data frame generated by the local sensor node and the target sequence number as the sequence number offset. The microcontrollers of the remaining sensing nodes subtract the sequence number offset from the write pointer index corresponding to the latest written data frame and add the physical slot depth of the local first-in-first-out circular buffer. Then, they perform a modulo operation on the physical slot depth to directly calculate and obtain the absolute storage index of the target data frame in memory. Under the condition that the serial number offset is greater than or equal to zero and less than the physical slot depth, the microcontrollers of the remaining sensing nodes generate a directly addressable physical address based on the absolute storage index and the memory base address, access the physical address, and forcibly overwrite the validity status flag bit of the located data frame to an invalid state.

9. The method for testing the sound insulation performance of a building according to claim 1, characterized in that, The specific steps of the sensing node performing logarithmic integral calculation of acoustic energy on the data frame that maintains an effective state, and transmitting equivalent continuous sound pressure level data when the lower limit of the test duration is reached, include: When the data frame maintaining the valid state is dequeued, the microcontroller of the sensing node maintains two global accumulators in parallel. One is used to calculate the sum of the effective signal energy of the data frame maintaining the valid state to obtain the total effective transmitted sound energy accumulation value, and the other is used to count the total number of data frames maintaining the valid state that actually participate in the integration. The microcontroller independently monitors the total number of data frames that actually participate in the integration and maintain an effective state. When the total number reaches the target effective frame number threshold determined by the lower limit of the test duration, the microcontroller stops data acquisition and data frame generation. The microcontroller calculates and obtains the equivalent time-averaged sound energy as the equivalent continuous sound pressure level data by dividing the total effective transmitted sound energy by the total number of data frames that actually participate in the integration and maintain the effective state. The microcontroller encapsulates the equivalent time-averaged acoustic energy into a result data packet and sends it back to the master control terminal. After confirming successful transmission, it cuts off the power supply to unnecessary peripherals and performs network sleep control.

10. A multi-point automatic measurement system for building sound insulation, characterized in that, A method for testing the sound insulation performance of a building as described in any one of claims 1-9, comprising: A main control terminal deployed inside the sound source room of the target building and a receiving node array deployed inside the receiving room of the target building; The main control unit includes an industrial computer, a power amplifier, a speaker, and a main communication module; The industrial control computer is connected to the power amplifier, the power amplifier is electrically connected to the speaker, and the main communication module is connected to the industrial control computer. The receiving node array includes multiple sensing nodes configured with globally unique integer identifiers; The sensing node includes a microphone, an analog-to-digital converter, a microcontroller, a memory, and a node communication module; The microphone is connected to the analog-to-digital converter, the analog-to-digital converter is connected to the microcontroller, and the microcontroller has a built-in hardware floating-point unit and is connected to the internal first-in-first-out circular buffer and the node communication module. The node communication module is used to perform wireless radio frequency communication with the main communication module and the node communication modules of the other sensing nodes.

Images