Resource scheduling optimization method for open RAN

Abstract

The invention relates to the field of construction of a new-generation mobile communication core network and an access network, and discloses a resource scheduling optimization method for an open RAN (Radio Access Network), which comprises the following steps that: a distribution unit determines the time offset of a subframe to be scheduled relative to a channel sampling moment, acquires the round-trip delay of an open forward transmission interface and calculates the delay fluctuation quantity; and determining an adjustment weight based on the time delay fluctuation quantity, performing attenuation correction on the channel prediction compensation bias, determining a modulation coding correction amplitude, and performing step-back processing on the modulation coding order when the feedback failure rate exceeds a protection threshold. An interaction mechanism that local short loop feedback executes correction on far-end long loop prediction is constructed, scheduling misalignment caused by forward jitter is effectively counteracted, and the stability and throughput of resource allocation under an open architecture are improved.

Description

Technical Field

[0001] This invention relates to the field of next-generation mobile communication core network and access network construction technology, and more specifically, to a resource scheduling optimization method for open RAN. Background Technology

[0002] In the current construction of next-generation mobile communication access networks, the open radio access network architecture adopts a near real-time radio access network intelligent controller to perform radio resource scheduling, thereby achieving decoupling and intelligentization of access network logical functions. The architecture deploys the scheduling decision plane and the physical layer execution plane in different logical entities, and the two transmit signaling through a standard interface. In actual operation, the radio access network intelligent controller determines the modulation and coding scheme based on the collected channel state information and sends scheduling instructions to the distribution unit through the standard interface. Due to the physical transmission delay and signaling processing time of the standard interface, there is a time offset relative to the channel sampling time when the scheduling instructions arrive at the physical layer execution terminal.

[0003] In high-speed mobile or multipath fading environments, channel states exhibit non-stationary time-varying characteristics. This time offset causes the channel state information relied upon by the distributed unit to degrade in timeliness when executing scheduling commands, resulting in a mismatch between scheduling decisions and instantaneous channel carrying capacity. Industry attempts have attempted to alleviate timeliness pressures by shortening the sampling period, but this approach leads to a non-linear increase in interface signaling overhead and cannot eliminate the inherent deterministic delay generated by the interface transmission link, resulting in an irreconcilable technical constraint between decoupling flexibility and scheduling determinism. In addition to the inherent transmission link constraints of the physical interface, existing scheduling control schemes have shortcomings in adapting to dynamic interface fluctuations at the algorithm level. For example, Chinese invention patent application CN118265091A discloses a joint function splitting and resource scheduling system for open wireless access networks. It uses a neural network model to coordinate function splitting and resource allocation on a large time scale. The decision logic is biased towards the remote intelligent controller. It does not consider the non-steady-state delay jitter caused by the high-concurrency traffic carried by the fronthaul interface. Due to the lack of real-time correction and adaptive protection mechanisms on the distributed unit side, the long loop prediction mode faces transient interface congestion or high-speed movement scenarios. The instruction issuance lag causes the scheduling bias to be inaccurate, which leads to the failure of nonlinear demodulation of the modulation and coding scheme and makes it difficult to maintain steady-state transmission performance under complex working conditions.

[0004] Therefore, how to utilize interface transmission characteristics and channel fading trajectories to construct an adaptive hedging mechanism to achieve accurate compensation for physical layer scheduling deviations under decoupled architecture constraints has become the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a resource scheduling optimization method for open RAN, comprising the following steps:

[0006] Step S1: The distributed unit parses the timestamp information in the Open Fronthaul Interface Protocol message to determine the time offset of the current subframe to be scheduled relative to the sampling time of the channel sounding signal.

[0007] Step S2: The distributed unit monitors the round-trip delay of the open fronthaul interface and calculates the delay fluctuation based on the root mean square difference of the round-trip delay within adjacent statistical periods.

[0008] Step S3: The distributed unit obtains the channel prediction compensation bias sent by the remote controller, and determines the adjustment weight based on the delay fluctuation calculated in step S2. The adjustment weight is used to perform attenuation correction on the channel prediction compensation bias in order to determine the modulation and coding strategy correction magnitude of the current subframe to be scheduled.

[0009] Step S4: The distributed unit obtains the hybrid automatic repeat request feedback status of the physical layer and calculates the feedback failure rate based on the number of negative response frames in the statistical window.

[0010] Step S5: When the feedback failure rate exceeds the protection threshold, the distributed unit performs a scheduling protection action. While maintaining the total amount of physical resource blocks allocated unchanged, it performs forced step-back processing on the modulation and coding level of subsequent downlink subframes by adding a negative offset to the modulation and coding order in the local scheduler, so as to offset the prediction compensation inaccuracy caused by the transient jitter of the round-trip delay. The final value of the modulation and coding level is controlled by the sum of the modulation and coding strategy correction magnitude and the negative offset.

[0011] Preferably, in step S3, the adjustment weight is determined as follows: the value of the adjustment weight is positively correlated with the delay fluctuation; wherein, the modulation and coding strategy correction amplitude is the product of the channel prediction compensation offset and the adjustment weight, and the modulation and coding strategy correction amplitude increases with the increase of the time offset, so as to improve the anti-interference margin of the physical layer transmission when the time offset increases.

[0012] Preferably, the method further includes the following steps: Step S6, the distribution unit monitors the fluctuation range of the round-trip delay and determines the allocation topology of the physical resource block according to the fluctuation range; Step S7, when the time delay fluctuation exceeds the coherence threshold, the distribution unit switches the mapping mode of the physical resource block from continuous allocation to discrete allocation, and the frequency hopping interval in the discrete allocation increases with the increase of the time delay fluctuation, and uses frequency domain diversity gain to offset the degradation of frequency domain channel characteristics caused by time offset.

[0013] Preferably, in step S5, the triggering logic for the scheduling protection action further includes: the distributed unit obtains the current remaining buffer depth of the open fronthaul interface; when the current remaining buffer depth is lower than the preset safety level of 10% and the feedback failure rate shows an increasing slope, the distributed unit determines that the scheduling command issued by the remote controller is invalid and switches to the short loop feedback scheduling mode based on the local channel quality indication of the distributed unit.

[0014] Preferably, in step S1, the path for determining the time offset is as follows: the distribution unit extracts the starting sampling time of the channel sounding reference signal; obtains the expected start time of the current downlink subframe to be scheduled in the air interface transmission, calculates the difference between the expected start time and the starting sampling time, and obtains the time offset.

[0015] Preferably, the modulation and coding strategy correction magnitude ΔMCS is determined by the following logic: ΔMCS=f(ΔT)⋅ω(∇D), where ΔT is the time offset, f(ΔT) is the attenuation function of the channel characteristics over time, ∇D is the delay fluctuation, and ω(∇D) is the adjustment weight.

[0016] Preferably, in step S5, the forced step-back process is executed as follows: the distribution unit establishes the index offset of the modulation and coding scheme lookup table in the local media access control layer scheduler; based on the real-time value of the feedback failure rate, the step value of the index offset is increased so that the modulation and coding level actually used in the physical downlink shared channel is lower than the nominal level in the remote controller instruction.

[0017] Preferably, in step S7, the switching logic of the mapping mode is as follows: the distribution unit determines the current channel coherence bandwidth and determines the subcarrier distribution step size of the resource block group in the discrete allocation according to the channel coherence bandwidth; when the jitter of the round-trip delay increases, the distribution unit reduces the length of the continuous resource block in a single scheduling and increases the mapping dispersion of the resource block in the entire system bandwidth.

[0018] Preferably, the logic for obtaining the time delay fluctuation is as follows: the distribution unit records the round-trip time delay data sequence of the open fronthaul interface within 20 consecutive scheduling cycles; the round-trip time delay data sequence is processed by a sliding window weighted average, the absolute value of the deviation between two adjacent statistical windows is calculated, and the absolute value of the deviation is determined as the time delay fluctuation.

[0019] Preferably, the method further includes the following steps: Step S8, the distributed unit asynchronously uploads the feedback failure rate and the execution status data of the forced step-back process to the remote controller through the reverse control link of the open interface, as the input feature for the remote controller to perform offline parameter calibration of the channel prediction compensation bias.

[0020] The embodiments of the present invention have at least the following beneficial effects:

[0021] 1. In open RAN resource scheduling, by extracting the interface transmission delay characteristics between the radio access network intelligent controller and the distributed unit, and combining the fading slope vector determined by time series analysis, an error hedging mechanism is established in the scheduling decision-making stage to address the architecture decoupling deviation. The inherent lag generated by control plane signaling transmission is transformed into adaptive bias parameters in the physical layer execution stage, eliminating the time offset between control plane command issuance and the instantaneous state of the physical layer channel. In the case of rapid channel fading, the system predicts the cumulative error within the transmission window based on the fading slope vector and performs nonlinear order reduction correction of the modulation and coding scheme. This avoids demodulation failure caused by the timeliness degradation of channel state information in the transmission path. Without changing the existing air interface protocol, the physical and logical alignment of scheduling decisions and execution states is achieved.

[0022] 2. By constructing a fading-sensitive factor using the Doppler frequency shift characteristics of the target user, this invention achieves precise adaptation of scheduling resources in the spatial dimension and mobility characteristics. By introducing the spatial correlation weight associated with Doppler frequency shift into the scheduling bias calculation logic, the system generates personalized compensation gains based on the differences in user mobility trajectories. In high-speed mobility scenarios, the channel decorrelation risk caused by Doppler frequency shift is suppressed by increasing the correction depth of the modulation and coding scheme. In low-speed or stationary scenarios, overprotected spectrum resources are released through attenuation compensation weights. This refined resource mapping based on physical motion states eliminates the resource waste of traditional single bias strategies in heterogeneous mobility environments and improves the stability of scheduling commands and spectrum utilization under complex operating conditions.

[0023] 3. Combining the hybrid automatic repeat request feedback response status on the distributed unit side, this invention constructs a local protection gating system for sudden interface jitter. When the distributed unit detects a step increase in the feedback failure gradient in real time, it determines that the prediction compensation of the remote controller is inaccurate due to sudden factors and triggers local step-back control. This mechanism, while maintaining the physical resource block quota unchanged, directly performs forced step-back adjustment on the modulation and coding order of subsequent radio frames on the distributed unit side, thereby smoothing the residual deviation caused by interface congestion in real time. This coordinated operation of short-loop feedback correction and long-loop feedforward prediction solves the phase mismatch problem between long-period feedback of the control plane and instantaneous changes in the physical layer, and enhances the steady-state transmission performance of the open architecture under extreme interference environments. Attached Figure Description

[0024] Figure 1 This is a flowchart of the resource scheduling method based on delay fluctuation correction and scheduling protection of the present invention;

[0025] Figure 2 This invention provides a resource scheduling logic architecture and parameter mapping diagram that integrates a local correction mechanism. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0027] The principles and spirit of the invention will be described with reference to several exemplary embodiments. It should be understood that these embodiments are given only to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. On the contrary, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0028] This invention provides a resource scheduling optimization method for Open RAN, based on a decoupled architecture consisting of a near-real-time RAN intelligent controller and a distributed unit (DMU). The near-real-time RAN intelligent controller and the DMU interact through an open fronthaul interface. To address the scheduling timeliness degradation problem caused by interface transmission in the decoupled architecture, the DMU executes a time alignment procedure to quantify the phase deviation of the scheduling decision relative to the instantaneous channel capability. During the execution of the time alignment procedure, the DMU parses the timestamp information in the open fronthaul interface protocol message to determine the time of the current subframe to be scheduled relative to the channel sounding signal sampling time. The distributed unit extracts the initial sampling time of the channel sounding reference signal and obtains the expected start time of the current downlink subframe to be scheduled in the air interface. The time offset ΔT is obtained by calculating the difference between the expected start time and the initial sampling time. Based on the non-steady-state characteristics of the delay generated by the fronthaul interface when carrying dynamic load, the distributed unit monitors the round-trip delay of the open fronthaul interface and extracts the link jitter characteristics to determine the delay fluctuation. The distributed unit records the round-trip delay data sequence of the open fronthaul interface in 20 consecutive scheduling cycles and calculates the delay fluctuation ∇D based on the root mean square difference of the round-trip delay data sequences in adjacent statistical cycles.

[0029] To achieve pre-correction of modulation and coding strategy under time-varying fading environment, the distributed unit obtains the channel prediction compensation offset sent by the remote controller and determines the adjustment weight based on the delay fluctuation ∇D. The distributed unit uses the adjustment weight to perform attenuation correction on the channel prediction compensation offset to determine the modulation and coding strategy correction magnitude ΔMCS of the current subframe to be scheduled. The calculation formula for the modulation and coding strategy correction magnitude ΔMCS is as follows: ΔMCS=f(ΔT)⋅ω(∇D), where ΔMCS is the modulation and coding strategy correction magnitude, ΔT is the time offset, f(ΔT) is the attenuation function of channel characteristics changing with time, ∇D is the delay fluctuation, and ω(∇D) is the adjustment weight, and the amount of adjustment weight ω(∇D) is... The value is positively correlated with the latency fluctuation ∇D; for scheduling inaccuracies caused by sudden congestion in the fronthaul link, the distribution unit performs local correction through local success rate gradient gating; the distribution unit monitors the hybrid automatic repeat request feedback status of the physical layer in real time and calculates the feedback failure rate based on the number of negative response frames in the statistical window; when the feedback failure rate exceeds the preset protection threshold of 10% and the remaining buffer depth on the current distribution unit side is detected to be lower than the preset safety level of 10%, the distribution unit determines that the scheduling instruction issued by the remote controller is invalid, and while maintaining the total amount of physical resource block allocation unchanged, it uses the local scheduler to add a negative offset in the modulation and coding level of the subsequent downlink subframe to perform forced step-back processing.

[0030] Considering the coherence degradation of the channel in the frequency domain, the distribution unit dynamically adjusts the allocation topology of physical resource blocks based on the delay fluctuation ∇D. When the delay fluctuation ∇D exceeds a preset coherence threshold, the distribution unit switches the mapping method of physical resource blocks from continuous allocation to discrete allocation, and determines the subcarrier distribution step size of the resource block group in the discrete allocation based on the channel coherence bandwidth of the current cell. This causes the frequency hopping interval in the discrete allocation to increase with the increase of the delay fluctuation ∇D, using frequency domain diversity gain to offset the frequency domain state inaccuracy caused by architecture delay. In addition, to address... To assess the impact of load fluctuations on scheduling accuracy, the distributed unit (DMU) establishes a linkage logic between computing load and air interface resource scheduling. The DMU periodically sends 64-byte probe messages through the management plane of the open fronthaul interface, recording the complete return time from message transmission to receiving the controller's response signal. This time is subtracted from the 0.1ms fixed bidirectional physical propagation delay calculated based on a 10km fiber optic cable length. The remaining delay difference is identified as the controller's processing time. The DMU monitors the execution time of the scheduling program within the near real-time wireless access network intelligent controller. The system determines the scheduling confidence index C by combining the execution time and the round-trip delay of the open fronthaul interface. When the scheduling confidence index C falls below the preset safety threshold due to controller overload, the system automatically reduces the effective weight of the modulation and coding bias parameters and increases the number of repeated transmissions of air interface control channel information. At the same time, the distributed unit asynchronously uploads the execution status data of the feedback failure rate and forced step-back processing to the remote controller through the reverse control link as a reference feature for execution parameter calibration.

[0031] Example 1: When the system is deployed in a high-density urban hotspot area and the open fronthaul interface carries concurrent traffic for multiple services, the open fronthaul interface generates non-stationary transmission delay fluctuations. At this time, the distributed unit parses the timestamp information in the protocol message and determines that the current subframe to be scheduled has a time offset ΔT of 4.5ms relative to the sampling time of the channel sounding signal. In order to correct the degradation of channel state information caused by the time offset ΔT, the distributed unit records the round-trip delay data sequence of the open fronthaul interface in 20 consecutive scheduling cycles and calculates the root mean square difference of the round-trip delay data sequence in adjacent statistical cycles, and determines that the delay fluctuation of the current link ∇D is 1.2ms. Based on this, the distributed unit determines an adjustment weight ω(∇D) of 0.75 according to the preset mapping procedure, and uses the adjustment weight ω(∇D) to perform attenuation correction on the channel prediction compensation bias issued by the near real-time wireless access network intelligent controller.

[0032] In this embodiment, when the function value corresponding to f(4.5ms) is 2.0 and ω(1.2ms) is 0.75, the distribution unit calculates the modulation and coding strategy correction amplitude ΔMCS of 1.5 according to the aforementioned calculation formula for the modulation and coding strategy correction amplitude ΔMCS. After receiving the modulation and coding strategy correction amplitude ΔMCS, the physical layer scheduler corrects the modulation and coding order from 28 to 26. By performing dynamic step-back processing on the modulation and coding scheme, the inaccuracy of the prediction model caused by delay jitter is offset, so that the demodulation signal-to-noise ratio margin of the physical downlink shared channel is maintained within the preset safe range, and the system maintains a hybrid automatic repeat request feedback failure rate of less than 10% under the condition of interface delay fluctuation.

[0033] Example 2: The experimental verification was built on a verification platform simulating an open wireless access network architecture, using a near real-time wireless access network intelligent controller and distributed units interconnected via an open fronthaul interface. The experiment employed Monte Carlo simulation to model the resource scheduling process under Rayleigh fading channels. To simulate signal interference in high-load urban environments, Gaussian white noise with a signal-to-noise ratio of 20dB was actively superimposed on the experimental signal source. In the parameter settings, the sampling period of the channel sounding signal was selected as 5ms to balance the real-time performance of the physical layer feedback with the processing load of the computing unit. Simultaneously, the basic round-trip delay of the open fronthaul interface was set to 8.2ms. In the first group of experiments, the control group used a pre-controller-only approach... The proposed method uses a scheduling method where the sample group executes a weight correction process based on the delay fluctuation ∇D. When the interface load fluctuation causes a time offset ΔT of 4.5ms and the delay fluctuation ∇D is 1.2ms, the proposed method determines an adjustment weight ω(∇D) of 0.75, and calculates a modulation and coding strategy correction amplitude ΔMCS of 1.5. Measurement data shows that the failure rate of the hybrid automatic repeat request feedback generated by the control group in the fading environment increases to 14.2%, while the proposed method reduces the feedback failure rate to 8.6% by dynamically stepping back the physical layer modulation and coding order. This data feedback demonstrates the technical output of the local correction logic in compensating for interface transmission delay.

[0034] To verify the adaptability of the scheme under different jitter intensities, three intensity gradients were set for the delay fluctuation ∇D: 0.5ms, 1.2ms, and 2.5ms. In the sample of this invention, as the delay fluctuation ∇D increases, the determined adjustment weight ω(∇D) gradually decreases from 0.92 to 0.55. The corresponding modulation and coding strategy correction amplitude ΔMCS shows a non-linear growth trend and triggers the discrete allocation topology switching of physical resource blocks under the extreme jitter condition of 2.5ms. Measurement data shows that when the jitter intensity is at a low gradient of 0.5ms, the system throughput is maintained at 145.2Mbps. When the jitter reaches the boundary condition of 2.5ms, the sample of this invention uses frequency domain diversity gain to maintain the throughput at a level of 128.6Mbps. Compared with the connection interruption phenomenon that occurred in the control group, the experimental results confirm the supporting role of the performance correction mechanism in system stability.

[0035] Example 3: When the system is applied to a vehicle-mounted mobile terminal operating in a highway environment with a speed of 120 km / h, the distributed unit extracts the Doppler frequency shift of the target user through physical layer detection signals. Furthermore, the time offset ΔT introduced by the open fronthaul interface is 6.0 ms, and the distributed unit is based on the Doppler frequency shift. The product of the time offset ΔT and the time offset quantizes the degree of channel timeliness degradation and determines the compensation depth of the prediction operator. To ensure the procedural and reproducible determination of scheduling parameters, the distributed unit executes the closed-loop operation of the prediction operator. The degradation function f(ΔT) follows the following calculation formula: Where f(ΔT) is the degradation function, and α is a preset frequency shift sensitivity coefficient with a value of 0.0005. The measured Doppler frequency shift is expressed in Hz, and ΔT is the time offset; when the measured Doppler frequency shift... At 400Hz, the distributed unit calculates the value of the degradation function f(ΔT) to be 2.2.

[0036] In the procedure for determining the modulation and coding strategy correction magnitude ΔMCS, the distributed unit performs a step mapping of the adjustment weight ω(∇D) based on the statistical variance of the round-trip delay over the past 100 scheduling cycles. When the statistical variance is within 0.5ms... 2 Up to 1.5ms 2 When the time interval is between 0.80 and 1.80, the weight ω(∇D) is adjusted to 0.80, while the measured statistical variance is 1.8ms. 2 And it exceeded 1.5ms 2 When determining the boundary, the weight ω(∇D) is adjusted to 0.65; the distributed unit calculates the modulation and coding strategy correction amplitude ΔMCS with an output value of 1.43, and uses the physical layer scheduler to adjust the modulation and coding order from 24 to 23. In response to the frequency domain coherence degradation in high-speed mobile environments, the distributed unit monitors the current delay fluctuation ∇D. When the delay fluctuation ∇D exceeds the coherence threshold of 2.0ms, the distributed unit switches the allocation topology of physical resource blocks from continuous mode to discrete mode. Based on the measured 1.2MHz channel coherence bandwidth, the distributed unit determines the step size of the resource block group in the discrete allocation to 4 physical resource blocks, thereby using frequency domain diversity gain to offset the execution lag loss caused by architecture decoupling.

[0037] Example 4: In a laboratory simulation environment, the system executes an offline calibration program for scheduling parameters. By constructing a channel sample library covering moving speeds from 5 km / h to 350 km / h and signal-to-noise ratios from 0 dB to 30 dB, the scheduling success rate under different round-trip delay jitter intensities is statistically analyzed to establish a mapping relationship between the delay fluctuation ∇D and the adjustment weight ω(∇D). The system uses the calculation formula ω(∇D)=exp(-β⋅∇D) to determine the weight distribution, where ω(∇D) is the adjustment weight, ∇D is the delay fluctuation, and β is an environmental sensitivity factor with a value of 0.12 in a line-of-sight transmission environment. In this way, the baseline data filling of the prediction operator is completed before formal deployment.

[0038] During the on-site deployment phase, the distributed unit performs a pre-calibration procedure by sending protocol messages of a preset length during idle time slots to measure the reference value of the round-trip delay of the physical link under different load pressures. Based on the length of the physical line and the routing level on site, it performs phase compensation correction on the initial zero point of the time offset ΔT. Simultaneously, it uses the measured channel fading envelope on site to perform step-by-step optimization on the frequency shift sensitivity coefficient α, so that the modulation and coding strategy correction amplitude ΔMCS is adapted to the electromagnetic wave propagation characteristics of the site, and finally establishes a parameter closed loop in the early stage of system operation.

[0039] Example 5: In a hybrid optoelectronic network environment with a fronthaul fiber optic cable length between 10km and 20km, the distribution unit executes a zero-point calibration procedure for physical propagation delay. The distribution unit sends an idle probe packet containing a high-precision timestamp to the near real-time wireless access network intelligent controller, measures the round-trip reference time of the signal in the physical medium, and calculates the link propagation delay using the signal propagation speed of 200,000 km / s in the optical fiber. When the measured round-trip reference time of the distributed unit is 0.15ms, the system sets this value as the physical starting zero point of the time offset ΔT, and uses the physical starting zero point to perform hardware-level synchronization alignment correction on the timestamp of the protocol message. During the service idle period, the distributed unit sends a hardware timestamp probe message to the near real-time wireless access network intelligent controller to measure the round-trip time of the message on the open fronthaul interface and subtract the preset internal processing time to determine the physical link propagation delay reference value. The reference value is used as the starting zero point of the time offset ΔT. The frequency shift sensitivity coefficient α in the degradation function f(ΔT) is determined by a step-by-step search procedure based on the block error rate feedback. α is selected to traverse in the range of 0.0001 to 0.0010 with a step size of 0.0001. The downlink physical shared channel block error rate is monitored under different step values. The value of the block error rate curve entering the stable range below 10% for the first time is determined as the frequency shift sensitivity coefficient α.

[0040] In determining the environmental sensitivity factor β, the distributed unit initiates an adaptive iterative process based on block error rate feedback. The distributed unit pre-sets a set of candidate factors with values ​​between 0.05 and 0.20 and a step size of 0.01. Using the weighting formula ω(∇D)=exp(-β⋅∇D), it performs periodic verification for each candidate value in the set for 500 radio frames. The distributed unit monitors the average block error rate change within the verification period. When the site environment is in a non-line-of-sight transmission state and the distributed unit detects that the physical downlink shared channel throughput reaches a local peak when β is 0.15, and the block error rate stabilizes below 0.1%, the system... This value is determined as the environmental sensitivity factor β of the current site. The distributed unit establishes a mapping relationship using the environmental sensitivity factor β. The value of the environmental sensitivity factor β depends on the electromagnetic wave propagation environment of the site. In the line-of-sight transmission environment, β is set to 0.12. In the non-line-of-sight transmission environment, the peak throughput of 1000 consecutive wireless frames is monitored. β is dynamically fine-tuned in the range of 0.05 to 0.20 with a step size of 0.01, so that the demodulation signal-to-noise ratio margin of the physical downlink shared channel is maintained in the preset safe range of 3dB to 5dB. When the output modulation and coding strategy correction amplitude ΔMCS is not an integer, the local scheduler rounds up to determine the final order offset and uses the physical layer to transmit anti-interference redundancy.

[0041] When the distributed unit monitors the execution time of the near real-time wireless access network intelligent controller When the value exhibits a monotonically increasing characteristic and reaches 15.0ms, the distributed unit initiates a scheduling quality degradation protection program, utilizing the execution time... With preset scheduling window duration The ratio determines the scheduling confidence index C; the distributed unit periodically reads the remaining space value of the local media access control layer buffer register. When the ratio of the remaining space value to the total register capacity is less than 10%, a buffer warning is triggered and the scheduling confidence index C is calculated simultaneously. The execution time of the near real-time wireless access network intelligent controller is obtained. Compared with the current scheduling cycle duration The ratio of the two values ​​yields the scheduling confidence index. ,in The duration for the controller to process scheduling instructions. To preset the wireless frame scheduling window width, when the scheduling confidence index C is below 0.7 and the feedback failure rate exhibits a slope-increasing characteristic, the distributed unit switches to a short-loop feedback mode based on local channel quality indicators. Downlink control information is retransmitted more frequently, and time diversity is used to offset downlink resource allocation inaccuracies caused by controller processing delay fluctuations. Under this condition, the distributed unit, based on the calculation results... The system determines the real-time weight of the current control plane signaling and uses the reciprocal of the scheduling confidence index C as a proportional operator to increase the number of repeated transmissions of downlink control information. At the same time, it reduces the effective weight of the modulation and coding bias parameter by 15%, thereby maintaining the transmission continuity of user plane data during the transient process of the controller handling drastic delay fluctuations.

[0042] The above description is only a few preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, technical solutions formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A resource scheduling optimization method for open RAN, characterized in that, Includes the following steps: Step S1: The distributed unit parses the timestamp information in the Open Fronthaul Interface Protocol message to determine the time offset of the current subframe to be scheduled relative to the sampling time of the channel sounding signal. Step S2: The distributed unit monitors the round-trip delay of the open fronthaul interface and calculates the delay fluctuation based on the root mean square difference of the round-trip delay within adjacent statistical periods. Step S3: The distributed unit obtains the channel prediction compensation bias sent by the remote controller, and determines the adjustment weight based on the delay fluctuation calculated in step S2. The adjustment weight is used to perform attenuation correction on the channel prediction compensation bias in order to determine the modulation and coding strategy correction magnitude of the current subframe to be scheduled. Step S4: The distributed unit obtains the hybrid automatic repeat request feedback status of the physical layer and calculates the feedback failure rate based on the number of negative response frames in the statistical window. Step S5: When the feedback failure rate exceeds the protection threshold, the distributed unit performs a scheduling protection action. While maintaining the total amount of physical resource blocks allocated unchanged, it performs forced step-back processing on the modulation and coding level of subsequent downlink subframes by adding a negative offset to the modulation and coding order in the local scheduler, so as to offset the prediction compensation inaccuracy caused by the transient jitter of the round-trip delay. The final value of the modulation and coding level is controlled by the sum of the modulation and coding strategy correction magnitude and the negative offset.

2. The resource scheduling optimization method for open RAN according to claim 1, characterized in that, In step S3, the adjustment weight is determined as follows: the value of the adjustment weight is positively correlated with the delay fluctuation; wherein, the modulation and coding strategy correction amplitude is the product of the channel prediction compensation offset and the adjustment weight, and the modulation and coding strategy correction amplitude increases with the increase of the time offset, so as to improve the anti-interference margin of the physical layer transmission when the time offset increases.

3. The resource scheduling optimization method for open RAN according to claim 1, characterized in that, Further steps include: Step S6: The distributed unit monitors the fluctuation range of the round-trip delay and determines the allocation topology of the physical resource block based on the fluctuation range. Step S7: When the time delay fluctuation exceeds the coherence threshold, the distributed unit switches the mapping mode of the physical resource block from continuous allocation to discrete allocation, and the frequency hopping interval in the discrete allocation increases with the increase of the time delay fluctuation, using frequency domain diversity gain to offset the degradation of frequency domain channel characteristics caused by the time offset.

4. The resource scheduling optimization method for open RAN according to claim 1, characterized in that, In step S5, the triggering logic for the scheduling protection action also includes: the distributed unit obtains the current remaining buffer depth of the open fronthaul interface; when the current remaining buffer depth is lower than the preset safety level of 10% and the feedback failure rate shows an increasing slope, the distributed unit determines that the scheduling command issued by the remote controller is invalid and switches to the short loop feedback scheduling mode based on the local channel quality indication of the distributed unit.

5. The resource scheduling optimization method for open RAN according to claim 1, characterized in that, In step S1, the path for determining the time offset is as follows: the distribution unit extracts the starting sampling time of the channel sounding reference signal; obtains the expected start time of the current downlink subframe to be scheduled in the air interface transmission, calculates the difference between the expected start time and the starting sampling time, and obtains the time offset.

6. The resource scheduling optimization method for open RAN according to claim 2, characterized in that, The modulation and coding strategy correction magnitude ΔMCS is determined by the following logic: ΔMCS=f(ΔT)⋅ω(∇D), where ΔT is the time offset, f(ΔT) is the attenuation function of the channel characteristics over time, ∇D is the delay fluctuation, and ω(∇D) is the adjustment weight.

7. The resource scheduling optimization method for open RAN according to claim 1, characterized in that, In step S5, the forced step-back process is executed as follows: the distributed unit establishes the index offset of the modulation and coding scheme lookup table in the local media access control layer scheduler; based on the real-time value of the feedback failure rate, the step value of the index offset is increased so that the modulation and coding level actually used in the physical downlink shared channel is lower than the nominal level in the remote controller instruction.

8. A resource scheduling optimization method for open RAN according to claim 3, characterized in that, In step S7, the switching logic of the mapping mode is as follows: the distribution unit determines the current channel coherence bandwidth and determines the subcarrier distribution step size of the resource block group in the discrete allocation according to the channel coherence bandwidth; when the round-trip delay jitter increases, the distribution unit reduces the length of the continuous resource block in a single scheduling and increases the mapping discreteness of the resource block in the entire system bandwidth.

9. A resource scheduling optimization method for open RAN according to claim 1, characterized in that, The logic for obtaining the time delay fluctuation is as follows: the distribution unit records the round-trip time delay data sequence of the open fronthaul interface within 20 consecutive scheduling cycles; the round-trip time delay data sequence is processed by sliding window weighted average, the absolute value of the deviation between two adjacent statistical windows is calculated, and the absolute value of the deviation is determined as the time delay fluctuation.

10. A resource scheduling optimization method for open RAN according to claim 1, characterized in that, It also includes the following steps: In step S8, the distributed unit asynchronously uploads the failure rate and execution status data of the forced step-back process to the remote controller via the reverse control link of the open interface, as the input feature for the remote controller to perform offline parameter calibration of the channel prediction compensation bias.

Images