A method and system for CPE signal adaptive enhancement and link switching

Abstract

The present application relates to the technical field of wireless communication and terminal network access control, and discloses a CPE signal adaptive enhancement and link switching method and system, wherein a link shadow fingerprint used for representing the fluctuation form of link quality in a short period is introduced, the combined features reflecting the short-time quality drop range, error code burst degree, time delay spike degree and apparent intensity level can be contained, and then the differentiation between the shielding shadow and the multipath false image in the scenario where the steel structure shielding and strong reflection coexist can be supported, a small amount of key link observation data is organized into a fingerprint type representation with scene recognition ability under the premise of not relying on multi-source data, and the basis for subsequent enhancement and switching control is provided. Thus, the present application introduces the link shadow fingerprint, makes the link control process have the identification ability for the special phenomenon that the apparent strong signal actually has error code burst, and makes the switching decision and the enhancement control form the collaborative management, so as to reduce the error switching, reduce the switching jitter and improve the service continuity.

Description

Technical Field

[0001] This invention relates to the field of wireless communication and terminal network access control technology, and in particular to a method and system for adaptive enhancement of CPE signals and link switching. Background Technology

[0002] In the engineering applications of mobile communication networks, CPEs are often deployed in complex operating environments with both strong obstruction and strong reflection to provide stable data backhaul links for field equipment. For example, in port remote-controlled cranes, unmanned truck platooning areas, or large steel structure storage areas, the site features large-scale metal components, continuously moving containers, and a dense distribution of multipath reflective surfaces. In such environments, CPEs may experience alternating periods of sudden drops in link quality due to obstruction and sudden spikes in bit errors due to multipath reflection, and these alternations are characterized by their abruptness, short periodicity, and unpredictability.

[0003] Existing link switching and enhancement strategies mostly rely on received power thresholds, average signal-to-interference-plus-noise ratios (SNR), or fixed-duration moving average quality metrics as triggering criteria, supplemented by simple hysteresis or hold times to suppress frequent switching. While these methods work in normal scenarios, in the aforementioned application scenario where steel structure obstruction and strong reflections coexist, the following phenomena easily occur: On the one hand, multipath reflections can make the received power of the reference signal appear very high, causing traditional methods to misjudge the link as good, thus missing the window for dealing with sudden bit error bursts and delay spikes; on the other hand, when obstruction and shadows cause a sharp deterioration in link quality, traditional methods may frequently trigger switching within a short period, leading to switching congestion, amplified link jitter, and significant service-side lag.

[0004] Therefore, in application scenarios where steel structure obstruction and strong reflection coexist, and services are sensitive to low latency and low bit error bursts, how to construct a link state representation that can distinguish between obstruction shadows and multipath artifacts, and thereby achieve coordinated control of signal enhancement and link switching to avoid misjudgment and incorrect switching and improve link stability, remains a technical problem that urgently needs to be solved in the field of CPE link control. Summary of the Invention

[0005] The purpose of this invention is to provide a CPE signal adaptive enhancement and link switching method and system, so as to at least solve the problems that traditional methods are easily misled by apparent strong signals, have difficulty in timely identifying sudden bit errors and delay spikes, and cause switching jitter and service interruption in special application scenarios where steel structure obstruction and strong reflection coexist.

[0006] To achieve the above objectives, the present invention provides a method for adaptive enhancement and link switching of CPE signals, the method comprising the following steps:

[0007] The system collects the link identifier, reference signal received power, signal-to-interference-plus-noise ratio, and block error rate of the serving link and candidate links, and generates round-trip delay for the serving link to form a link observation record.

[0008] A short window sequence is formed based on the link observation records, and the link shadow fingerprint is extracted from the short window sequence to characterize the quality fluctuation, block bursts and delay spikes of the current link in a short period of time.

[0009] Using the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input, change point detection is performed to generate shadow event segment flags and shadow intensity;

[0010] Using the link shadow fingerprint, the shadow event segment flag, and the shadow intensity as observations, perform link state discrimination to output the probability of normal state, the probability of occlusion shadow state, the probability of multipath artifact state, and the current state;

[0011] Based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block error rate of the candidate links, signal enhancement control and link switching control are performed.

[0012] Optionally, the link identifier, reference signal received power, signal-to-interference-plus-noise ratio (SINNR), and block error rate of the serving link and candidate links are collected, and the round-trip time delay of the serving link is generated to form a link observation record. This process specifically includes:

[0013] Obtain the link identifiers corresponding to the serving link and the candidate link, and read the corresponding reference signal received power, signal-to-interference-plus-noise ratio and block error rate respectively;

[0014] Perform a lightweight probe on the control server to obtain the round-trip latency of the service link;

[0015] The link identifier, reference signal received power, signal-to-interference-plus-noise ratio, block bit error rate, and round-trip time are written into the same link observation record at the same sampling time.

[0016] Optionally, the steps of forming a short-window sequence based on the link observation records and extracting a link shadow fingerprint from the short-window sequence to characterize the quality fluctuations, block bursts, and delay spikes of the current link within a short period of time specifically include:

[0017] Extract the most recent consecutive records of a preset window length from the link observation records to form a short window sequence;

[0018] Within the short window sequence, the signal-to-interference-plus-noise ratio fluctuation amplitude, the proportion of block error rate exceeding the preset error threshold, the maximum jump amplitude of adjacent samples of round-trip delay, and the average value of the reference signal received power of the statistical service link are obtained to obtain quality fluctuation characteristics for characterizing the degree of short-term quality degradation, error block burst characteristics for characterizing the degree of error block burst, delay spike characteristics for characterizing the degree of delay spike, and strength mean characteristics for characterizing the apparent signal strength level.

[0019] The quality fluctuation feature, the block burst feature, the delay spike feature, and the intensity mean feature are combined to form a link shadow fingerprint.

[0020] Optionally, using the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input, a step of change point detection to generate shadow event segment flags and shadow intensity is performed, specifically including:

[0021] Based on the historical signal-to-interference-plus-noise ratio (SIR) sequence of the service links in the link observation records, a baseline mean SIR is generated, and the cumulative offset statistic of SIR is updated based on the baseline mean SIR and the current SIR in the link shadow fingerprint.

[0022] Based on the historical round-trip delay sequence of the service link in the link observation record, a baseline mean of round-trip delay is generated, and based on the baseline mean and the current round-trip delay in the link shadow fingerprint, the cumulative offset statistic of round-trip delay is updated.

[0023] When the cumulative offset statistic of the signal-to-interference-plus-noise ratio or the cumulative offset statistic of the round-trip delay exceeds the corresponding threshold, the shadow event segment flag is determined to be valid, and the shadow intensity is generated based on the relative relationship between the cumulative offset statistic and the threshold.

[0024] Optionally, using the link shadow fingerprint, the shadow event segment flag, and the shadow intensity as observations, a link state discrimination step is performed to output the probability of normal state, the probability of occlusion shadow state, the probability of multipath artifact state, and the current state. This step specifically includes:

[0025] The link state is divided into a set of states: normal state, occlusion shadow state, and multipath artifact state. An observation sequence is constructed based on the link shadow fingerprint, shadow event segment flag, and shadow intensity.

[0026] The Hidden Markov Model is invoked to perform inference based on the observation sequence, and the state probability corresponding to each state is output. The state with the highest state probability is taken as the current state.

[0027] Optionally, based on the current state, the probabilities of each state, and the shadow intensity, signal enhancement control is performed, specifically including:

[0028] When the current state is occlusion shadow state and the probability of occlusion shadow state reaches the preset condition, the occlusion enhancement control mode is triggered. The receiving gain target is increased or the diversity union priority is increased based on the shadow intensity, so as to improve the signal-to-interference-plus-noise ratio of subsequent sampling and reduce block bursts.

[0029] When the current state is a multipath illusion state and the probability of the multipath illusion state reaches the preset condition, the illusion suppression control mode is triggered to tighten the maximum gain limit or suppress the rapid rise of gain, thereby increasing the sensitivity to block error rate and round-trip delay anomalies and reducing the misleading effect of apparent strong signals on control decisions.

[0030] If the triggering conditions for the occlusion enhancement control mode and the artifact suppression control mode are not met, the normal enhancement control mode is triggered.

[0031] Optionally, based on the reference signal received power, signal-to-interference-plus-noise ratio, and block bit error rate of the candidate link, a link switching control step is performed, specifically including:

[0032] For each serving link and each candidate link, a candidate link score and a serving link score are generated based on the degree of improvement in signal-to-interference-plus-noise ratio, the degree of reduction in block bit error rate, and the level of received power of reference signal.

[0033] The probability of occlusion and shadow states is used to increase the influence of signal-to-interference-plus-noise ratio on the candidate link score, and the probability of multipath artifact states is used to reduce the influence of reference signal received power on the candidate link score and increase the penalty of block error rate on the candidate link score.

[0034] Based on the candidate link score and the service link score, read the historical failure count or risk count corresponding to the candidate link in the link fingerprint cache table, and use the failure count or risk count as a penalty factor for the link score to suppress repeated selection of high-risk links. Determine the candidate link with the largest link score after penalty as the target link, and calculate the score advantage between the target link and the service link.

[0035] Based on shadow intensity and state probability, an adaptive switching threshold is generated and a link switching is triggered.

[0036] Optionally, based on shadow intensity and state probability, an adaptive switching threshold is generated and a link switching step is triggered, specifically including:

[0037] An adaptive handover threshold is generated based on shadow intensity, occlusion shadow state probability, and multipath artifact state probability to trigger link handover.

[0038] The target link's score advantage is compared with the adaptive handover threshold to generate a candidate trigger flag when the score advantage exceeds the adaptive handover threshold.

[0039] A continuity check is performed on the candidate trigger flags to generate a switching trigger flag when the candidate trigger flags continuously meet a preset number of times, and then a link switch is performed.

[0040] Optionally, a link switching procedure is performed, specifically including:

[0041] When the switching trigger flag is valid, a switching transaction form is generated, and the trigger time, service link identifier, target link identifier, scoring advantage quantity, adaptive switching threshold, link shadow fingerprint, shadow intensity, current state and probability of each state are written into the switching transaction form;

[0042] Based on the switching transaction form, the default data plane exit is switched from the service link to the target link, and after the switch, link observation records are continuously generated according to the preset verification window.

[0043] The round-trip time and block error rate of the target link are summarized in the verification window and compared with the baseline before handover recorded in the handover transaction form, so as to generate a handover success flag when the round-trip time decreases and the block error rate decreases to a preset level.

[0044] If the switch success flag does not meet the preset conditions, a rollback is executed to restore the service link, and the failure information is written to the link fingerprint cache table for the failure count or risk count corresponding to the target link.

[0045] Furthermore, to achieve the above objectives, the present invention also provides a CPE signal adaptive enhancement and link switching system, comprising:

[0046] The acquisition module is used to acquire the link identifier, reference signal received power, signal-to-interference-plus-noise ratio and block error rate of the serving link and candidate links, and to generate round-trip delay for the serving link to form a link observation record;

[0047] The extraction module is used to form a short window sequence based on the link observation records, and extract the link shadow fingerprint from the short window sequence to characterize the quality fluctuation, block bursts and delay spikes of the current link in a short period of time.

[0048] The generation module is used to perform change point detection to generate shadow event segment flags and shadow intensity by taking the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input;

[0049] The output module is used to perform link state discrimination based on the link shadow fingerprint, the shadow event segment flag and the shadow intensity as observations, and output the probability of normal state, the probability of occlusion shadow state and the probability of multipath artifact state and the current state.

[0050] The control module is used to perform signal enhancement control and link switching control based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block bit error rate of the candidate links.

[0051] The beneficial effects of this invention are as follows: It proposes a CPE signal adaptive enhancement and link switching method and system. By introducing a link shadow fingerprint to characterize the link quality fluctuation pattern within a short period, it can include combined features reflecting the short-term quality drop magnitude, the degree of bit error bursts, the degree of delay spikes, and the apparent strength level. This supports the differentiation between occlusion shadows and multipath artifacts in scenarios where steel structure occlusion and strong reflection coexist. This link shadow fingerprint can organize a small amount of key link observation data into a fingerprint-like representation with scene identification capabilities without relying on multi-source data, thus providing a basis for subsequent enhancement and switching control. Therefore, this invention enables the link control process to identify the special phenomenon of apparent strong signals but actual bit error bursts, and enables switching decisions and enhancement control to form collaborative management, thereby reducing false switching, reducing switching jitter, and improving service continuity. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the CPE signal adaptive enhancement and link switching method according to an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the CPE signal adaptive enhancement and link switching system according to an embodiment of the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0055] This invention provides a method for adaptive enhancement and link switching of CPE signals, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the CPE signal adaptive enhancement and link switching method according to an embodiment of the present invention.

[0056] In this embodiment, a CPE signal adaptive enhancement and link switching method is provided, the method comprising the following steps:

[0057] S1: Collect the link identifier, reference signal received power, signal-to-interference-plus-noise ratio and block error rate of the serving link and candidate links, and generate round-trip delay for the serving link to form a link observation record.

[0058] Specifically, step S1 involves collecting, aligning, and merging the link identifier, reference signal received power, signal-to-interference-plus-noise ratio, block error rate, and round-trip delay of the serving link and candidate links at the same sampling time to form a link observation record sequence.

[0059] First, link identifiers and wireless quality metrics are collected and aligned to the sampling time. In this embodiment of the invention, at each sampling time, the link identifiers corresponding to the serving link and candidate links are read from the CPE wireless side statistics interface, and the reference signal received power, signal-to-interference-plus-noise ratio (SINNR), and block error rate (BER) of each link are read respectively. It is easy to understand that the link identifier is used to establish a one-to-one tracking relationship when multiple links coexist, the reference signal received power reflects the apparent coverage strength, the SINNR reflects the quality of the effective link, and the BER reflects the service availability risk.

[0060] It should be noted that in special application scenarios where steel structure obstruction and strong reflection coexist, multipath reflection may cause the reference signal received power to remain high, while the signal-to-interference-plus-noise ratio (SINR) may fluctuate significantly in a short period of time and the block error rate may suddenly increase. Therefore, in this embodiment of the invention, the above indicators are collected and aligned at the same sampling time so that abnormal combinations that appear strong but are actually poor can be identified in the future.

[0061] Then, the service link round-trip time (RTD) is generated and lightweight constraints are applied. In this embodiment of the invention, lightweight probing is performed on the service link at the sampling time to obtain the RTD, and it is merged with the collected wireless quality indicators under the same sampling sequence number and written into the link observation record. It is easy to understand that RTD can reflect the impact of retransmission, queuing, and scheduling degradation on services. When occlusion shadows or sudden bit errors occur, the RTD often exhibits spikes or continuous increases. Therefore, RTD can serve as service-side verification data for wireless side indicators.

[0062] In practical applications, round-trip time (RTT) can be obtained by sending probe messages and recording the sending time, and receiving return packets and recording the receiving time. It should be noted that, to avoid occasional measurement jitter from misleading subsequent detections, in one optional implementation, the RTT can be truncated with an upper limit or outlier replaced. For example, when the RTT exceeds a preset upper limit, the upper limit value can be used as a substitute, thereby ensuring the stability of subsequent statistics. However, this invention does not limit the use of this constraint method.

[0063] Finally, the link observation records are written to form a traceable sequence. In this embodiment of the invention, the obtained link identifier, reference signal received power, signal-to-interference-plus-noise ratio, block bit error rate, and round-trip time are written into the same link observation record, and a link observation record sequence is formed according to the sampling sequence number. It should be noted that this sequence can be stored in a circular buffer or queue structure so that subsequent continuous records can be quickly truncated according to the nearest window length. Furthermore, by binding a sampling sequence number or sampling timestamp to each record, the consistency of the order of subsequent short-window sliding updates can be guaranteed, so that subsequent statistical actions strictly correspond to the link state evolution over the most recent period.

[0064] S2: A short window sequence is formed based on the link observation records, and the link shadow fingerprint is extracted from the short window sequence to characterize the quality fluctuation, block bursts and delay spikes of the current link in a short period of time.

[0065] Specifically, step S2 constructs a short-window sequence based on the link observation record sequence formed in step S1, and statistically analyzes the signal-to-interference-plus-noise ratio fluctuations, block bursts, and round-trip delay spikes of the serving link within the short-window range, while retaining the apparent strength level of the reference signal received power. The above statistical results are then combined to form a link shadow fingerprint. It should be noted that the link shadow fingerprint is a specific data structure defined in this embodiment of the invention, used to characterize the morphological features of link quality within a short period. Its purpose is to enable subsequent modules to distinguish between occlusion shadows and multipath artifacts, two types of anomalies most easily confused in special scenarios.

[0066] First, a short window sequence is constructed and a sliding update is performed. In this embodiment of the invention, the most recent consecutive records of a preset window length are extracted from the link observation record sequence obtained in step S1 to form a short window sequence, and the short window sequence is slid-updated each time a new sample arrives.

[0067] It should be noted that the window length setting is used to strike a balance between response speed and jitter resistance: a window length that is too short may make the system sensitive to noise, while a window length that is too long may dilute sudden changes. In practical applications, the window length can be configured according to the sensitivity of the business, but this invention does not limit the specific value.

[0068] Then, the signal-to-interference-plus-noise ratio (SINR) short-window fluctuation characteristics, block error burst ratio characteristics, round-trip delay spike characteristics, and apparent strength characteristics of the received power of the reference signal are calculated.

[0069] (1) To address the short-window fluctuation characteristics of the signal-to-interference-plus-noise ratio (SINR), the peak-to-valley difference of the SINR of the service link is statistically analyzed within the short-window sequence to obtain fluctuation characteristics that reflect the magnitude and degree of short-term quality decline.

[0070] It should be noted that under occlusion and shadow, the signal-to-interference-plus-noise ratio (SNR) tends to drop rapidly and remain low, resulting in a larger peak-to-valley difference and a persistent trough. Under multipath artifacts, the SNR may fluctuate unstablely, potentially increasing the peak-to-valley difference as well, but this, combined with bursts of false blocks and delay spikes, presents a different fingerprint pattern. Furthermore, by using the peak-to-valley difference instead of the mean, short-term abrupt changes can be captured more sensitively, making it more suitable for special scenarios involving sudden occlusion and sudden reflections.

[0071] (2) Based on the characteristics of the proportion of block error bursts, the service link block error rate is judged by threshold within the short window sequence, and the proportion of exceeding the threshold is counted as the characteristics of block error bursts.

[0072] It should be noted that this threshold-crossing ratio directly reflects the persistence of sudden errors within the short window. One typical phenomenon of multipath artifacts is that the apparent strength is not low, but the erroneous blocks cross the threshold frequently. Therefore, this feature can play a key constraint role in subsequent state discrimination and handover control. In practical applications, the erroneous block threshold can be configured according to the service type. For example, a stricter threshold can be used for remote control services, but this invention does not limit the threshold configuration method.

[0073] (3) For the round-trip delay spike feature, perform adjacent difference on the service link round-trip delay within the short window sequence, and take the maximum value of the difference as the delay spike feature.

[0074] It should be noted that using maximum jumps instead of average changes can highlight the abnormal behavior of sudden stuttering, thus making it more sensitive to bit error retransmissions or sudden increases in queuing. Furthermore, in cases where retransmissions are caused by multipath artifacts, even if the received power of the reference signal is high, round-trip delay spikes can serve as key evidence in subsequent judgment and control.

[0075] (4) To obtain the apparent strength characteristics of the reference signal received power, the average of the reference signal received power of the serving link is calculated within the short window sequence.

[0076] It should be noted that this invention does not negate the engineering significance of the reference signal received power, but rather uses it as a comparative dimension of apparent coverage strength in conjunction with effective quality characteristics, thereby avoiding being misled by apparent strength in strong reflection scenarios. Furthermore, when the apparent strength is high but the characteristics of block bursts and delay spikes are also high, subsequent state determination can be more inclined towards multipath artifact states.

[0077] Finally, the link shadow fingerprint is formed by combining the features. In this embodiment of the invention, fluctuation features, block burst features, delay spike features, and apparent intensity features are combined to form the link shadow fingerprint. It should be noted that this combination is not a simple stacking, but is used to enable subsequent modules to identify short-term morphological patterns, thereby maintaining distinguishability even in special scenarios where occlusion shadows and multipath artifacts alternate. It is easy to understand that the link shadow fingerprint can be stored using structured fields or fixed-length vectors, but this invention does not limit its specific organization.

[0078] S3: Using the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input, perform change point detection to generate shadow event segment flags and shadow intensity.

[0079] Specifically, step S3 takes the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as inputs, performs change point detection to identify the intervals where the link quality undergoes continuous abrupt changes, and generates shadow event segment flags and shadow intensity. It should be noted that this embodiment introduces a CUSUM change point detection mechanism that is sensitive to continuous deviations but not sensitive to brief jitter, which can adapt to the identification of abnormal segments with continuous degradation characteristics such as occlusion shadows; at the same time, the CUSUM statistics combined with round-trip delay can provide mutual verification for service degradation caused by bit error retransmission, thereby improving the ability to capture multipath artifact-type abnormal segments.

[0080] First, a signal-to-interference-plus-noise ratio (SIR) baseline is constructed and the cumulative SIR offset statistic is updated. In this embodiment of the invention, the mean SIR baseline is constructed based on the historical SIR sequence, and its calculation can be expressed as:

[0081] ;

[0082] In the formula, This represents the baseline mean of the signal-to-interference-plus-noise ratio (SINR). Indicates the historical length used to calculate the baseline; Indicates the sampling sequence number; Indicates the service link; This represents the signal-to-interference-plus-noise ratio (SIR) of historical samples. It should be noted that the baseline is used to characterize the normal level; deviations of the current value from the baseline will persist when shadows are present.

[0083] Furthermore, the cumulative offset statistic of the signal-to-interference-plus-noise ratio (SIR) is updated based on the baseline mean and the current SIR, and its update can be expressed as:

[0084] ;

[0085] In the formula, This represents the cumulative offset statistic of the signal-to-interference-plus-noise ratio (SINR). Indicates the current sampling signal-to-interference-plus-noise ratio; The term represents the drift compensation; `max(·)` represents the maximum value operation. It should be noted that the drift compensation term is used to suppress meaningless accumulation caused by normal jitter, making the statistic more focused on continuous deterioration; the maximum value operation is used to limit the statistic to non-negative accumulation, avoiding short-term rebounds that could lead to negative accumulation and destroy the physical meaning of the cumulative deviation. In practical applications, when occlusion shadows cause a continuously low signal-to-interference-plus-noise ratio, the statistic will increase rapidly, thus providing a basis for the subsequent generation of anomaly segment markers.

[0086] Step S302: Construct a round-trip delay baseline and update the cumulative round-trip delay offset statistics. In this embodiment of the invention, the mean of the round-trip delay baseline is constructed based on the historical round-trip delay sequence, and its calculation can be expressed as:

[0087] ;

[0088] In the formula, This represents the baseline mean of round-trip delay; Indicates the historical length used to calculate the baseline; Indicates the sampling sequence number; Indicates the service link; This represents the historical round-trip time. It should be noted that when bit error retransmissions or queuing backlogs occur, the round-trip time will remain higher than the baseline. Therefore, the baseline can provide a reference for continuous degradation.

[0089] Furthermore, the cumulative offset statistic of the round-trip delay, updated based on the baseline mean and the current round-trip delay, can be expressed as:

[0090] ;

[0091] In the formula, This represents the cumulative offset statistic for round-trip delay; Indicates the current round-trip time for sampling; This represents the drift compensation term. It should be noted that the round-trip delay statistic is used to characterize the service degradation trend, especially when multipath artifacts cause retransmissions, it can provide strong evidence; at the same time, in conjunction with the signal-to-interference-plus-noise ratio statistic, it can avoid missing out on abnormal segments that appear strong but have degraded services by relying solely on radio-side indicators.

[0092] Step S303: Generate a shadow event segment flag and calculate the shadow intensity. In this embodiment of the invention, the shadow event segment flag is generated based on the comparison between the cumulative offset statistic of signal-to-interference-plus-noise ratio (SIR) and the cumulative offset statistic of round-trip delay (RTD) and their respective thresholds. This can be achieved by: flagging the shadow event segment when the cumulative offset statistic of SIR is greater than the corresponding SIR CUSUM threshold or the cumulative offset statistic of RTD is greater than the corresponding RTD CUSUM threshold. It should be noted that this flag is used to provide a discrete judgment on whether an abnormal segment has been entered, enabling subsequent state discrimination and control to have event boundary inputs.

[0093] Furthermore, to characterize the severity of the abnormal segment, this embodiment also calculates the shadow intensity, which can be expressed as:

[0094] ;

[0095] In the formula, Indicates shadow intensity; and represents the CUSUM thresholds for signal-to-interference-plus-noise ratio (SIR) and round-trip delay (RTD), respectively; min(·) and max(·) represent the minimum and maximum values, respectively. It should be noted that this calculation reflects the approximation degree through the ratio of the statistical quantity to the threshold, and upper and lower bounds ensure the stability and usability of the intensity value. It is easy to understand that a higher shadow intensity indicates a more significant and continuous deviation, and subsequent enhancement intensity and switching threshold can be adaptively adjusted accordingly. In practical applications, as occlusion gradually worsens, the shadow intensity will gradually increase, thereby driving the control strategy to evolve from observation and mild enhancement to more decisive switching.

[0096] S4: Using the link shadow fingerprint, the shadow event segment flag, and the shadow intensity as observations, perform link state discrimination to output the probability of normal state, the probability of occlusion shadow state, the probability of multipath artifact state, and the current state.

[0097] Specifically, step S4 uses the link shadow fingerprint, shadow event segment flag, and shadow intensity as observations to perform link state discrimination, outputting the probability of normal state, the probability of occlusion shadow state, and the probability of multipath artifact state, and determines the current state. It should be noted that this embodiment employs a Hidden Markov Model, utilizing state continuity to suppress misjudgments caused by short-term jitter, and outputs the state matching degree in probabilistic form, thus providing subsequent control not only with explicit mode selection but also with continuously adjustable probabilistic support.

[0098] First, observations are constructed and a state set is established. In this embodiment of the invention, a state set is established that includes at least a normal state, an occluded shadow state, and a multipath artifact state, and observations are constructed using link shadow fingerprints, shadow event segment flags, and shadow intensity. It should be noted that combining short-window morphological features and event segment information into observations allows the model to utilize both morphological evidence and abrupt boundary evidence, thus making it more suitable for special scenarios with alternating occlusion and reflection.

[0099] Then, the Hidden Markov Model (HMM) inference is performed and the state probabilities are output. In this embodiment of the invention, the observations are input into the forward inference process of the HMM to calculate the probabilities of each state. To clarify the execution method, the forward recursion can be expressed as:

[0100] ;

[0101] ;

[0102] In the formula, M represents the number of states; This represents the front vector in the m-th state when the sampling sequence number is k; This represents the transition probability from the j-th state to the m-th state; Represents the observable; This represents the emission probability of an observation occurring in the m-th state; This represents the normalized probability of the m-th state. It should be noted that the inference process of the above Hidden Markov Model is a relatively mature and directly applicable inference process in the existing technology, and will not be elaborated further in this embodiment.

[0103] In practical applications, transition probabilities are used to reflect the assumption of state continuity, so that occlusion shadows and multipath artifacts do not jump irregularly between adjacent samples, thereby suppressing frequent misjudgments caused by short-term fluctuations caused by strong reflections.

[0104] Finally, the current state is determined and output. In this embodiment of the invention, the state corresponding to the maximum state probability is determined as the current state. It should be noted that this embodiment outputs three types of state probabilities simultaneously, so that subsequent steps can continuously adjust the enhancement intensity, scoring weight, and switching threshold; it is easy to understand that the current state is used to select the control mode, and the state probability is used to determine the control intensity and caution level.

[0105] S5: Based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block error rate of the candidate links, perform signal enhancement control and link switching control.

[0106] Specifically, step S5 generates signal enhancement control and link switching control based on the current state, state probability, shadow intensity, and candidate link quality indicators; when the triggering conditions are met, a switching transaction record is executed and the link switching is completed; after switching, verification and failure rollback are performed, and the link fingerprint cache table is updated and fed back to the subsequent scoring and selection process, thus forming a feedback coordination. It should be noted that the key to this step is to solve the problem that traditional methods are unable to simultaneously cope with the rapid deterioration of occlusion and the misleading multipath artifacts in special scenarios.

[0107] First, enhancement control parameters are generated and applied to subsequent observations. In this embodiment of the invention, the enhancement control mode is selected and enhancement control parameters are generated based on the current state and state probability. It should be noted that when the tendency for occlusion and shadowing is significant, the enhancement control biases to increase the effective operating range of the receiver; when the tendency for multipath artifacts is significant, the enhancement control biases to suppress gain overshoot and interference amplification, thereby avoiding the counterproductive effect of increasing enhancement while making things worse.

[0108] In one feasible implementation, when the occlusion shadow tendency is significant, a target received intensity can be generated and the received gain adjusted accordingly. This can be achieved by accumulating the product of the intensity adjustment coefficient and the shadow intensity based on the regular target intensity. It should be noted that a higher shadow intensity indicates more severe occlusion, and the target received intensity should be increased accordingly, thereby driving the receiver to be more proactive towards weak signals. In practical applications, this target can be mapped to a target range or diversity of automatic gain control and prioritized, but this invention does not limit the specific hardware mapping method.

[0109] Furthermore, when multipath artifacts are significant, to suppress apparent strength misleading and gain overshoot, the maximum gain upper limit can be tightened. This can be achieved by subtracting the product of the tightening coefficient and the multipath artifact state probability from the normal maximum gain to obtain the tightened maximum gain. It should be noted that the higher the multipath artifact probability, the greater the tightening, thereby reducing the risk of amplifying both reflections and interference. It is easy to understand that the above enhancement control will affect the received power, signal-to-interference-plus-noise ratio, and block error rate of the reference signal obtained from subsequent sampling; therefore, the effect of the enhancement control will re-enter the link observation recording sequence through step S1.

[0110] Then, candidate link scores are calculated, buffer penalties are introduced, and target links are determined. In this embodiment of the invention, candidate link scores are generated based on the signal-to-interference-plus-noise ratio (SIR), block error rate (BER), and reference signal received power (RSP) of the candidate links, and target links are determined. Specifically, the score for each candidate link can be measured by calculating the weighted sum of the SIR, BER, and RSP of the candidate link at the sampling time with their corresponding dynamic weights. When performing the weighted summation, the SIR and RSP use their original values, and the BER uses a negative value. This score incorporates effective quality and service risk into the same evaluation framework, so that the selection of target links no longer depends solely on apparent strength.

[0111] Furthermore, to enable the scoring to adaptively adjust according to the anomaly type, this embodiment updates the dynamic weights based on the state probability. The update method can be the product of the base weights corresponding to the dynamic weights of the signal-to-interference-plus-noise ratio (SINR), block bit error rate (BER), and reference signal received power (RSP), respectively, and a real-time adjustment coefficient. Specifically, the real-time adjustment coefficient for SINR is 1 minus the product of a first preset coefficient and the multipath artifact state probability; the real-time adjustment coefficient for BER is 1 plus the product of a second preset coefficient and the multipath artifact state probability; and the real-time adjustment coefficient for RSP is 1 minus the product of a third preset coefficient and the occlusion / shadow state probability. It should be noted that when the multipath artifact probability increases, the RSP weight is reduced while the bit error rate penalty weight is increased, thereby avoiding misleading selection due to inflated apparent strength; when the occlusion / shadow probability increases, the SINR weight is increased, thus enabling faster selection of links with better effective quality.

[0112] In practical applications, to avoid repeatedly selecting links that are prone to failure in similar scenarios, this embodiment also introduces a risk penalty through a link fingerprint cache table. The cache table records the risk count of each candidate link, and a penalty score is calculated based on this risk count. This penalty score can be represented by subtracting the product of the penalty coefficient and the risk count from the calculated candidate link score.

[0113] Subsequently, the target link is selected as the candidate link with the highest penalized link score. The difference between the penalized score of the target link and the score of the serving link is then calculated to represent the score advantage of the target link relative to the serving link. It should be noted that this score advantage is used for subsequent threshold comparisons and is a direct input for handover triggering.

[0114] Next, an adaptive switching threshold and candidate trigger flags are generated, and continuity confirmation is performed. In this embodiment of the invention, an adaptive switching threshold is generated based on shadow intensity and state probability, and candidate trigger flags are generated by comparing the scoring advantage with the threshold. To clarify the execution method, the adaptive switching threshold can be expressed as:

[0115] ;

[0116] In the formula, H(k) represents the switching threshold; Indicates the basic threshold; a, b, c represent the threshold adjustment coefficients; Indicates the probability of shadow occlusion; This represents the probability of a multipath artifact state. It should be noted that increasing shadow intensity will increase the threshold urgency adjustment term, making the system more sensitive to continuous deterioration; increasing the probability of occlusion shadows will lower the threshold for more decisive switching; increasing the probability of multipath artifacts will raise the threshold for more cautious switching, thus suppressing false switching and jitter.

[0117] Furthermore, a count is performed when a candidate trigger flag meets the trigger condition (i.e., the score advantage is greater than the switching threshold). Considering that a single comparison may be affected by short-term fluctuations, this embodiment performs continuous cumulative counting on the candidate trigger flags. When the continuous count exceeds the hold count threshold, a switching trigger flag is generated. It should be noted that the continuity confirmation is used to filter false threshold crossings caused by short-term jitter, thereby significantly reducing the risk of frequent switching in scenarios where strong reflections cause index fluctuations. In practical applications, the hold count threshold can be configured according to business tolerance, but this invention does not limit the specific value.

[0118] Following this, a switching transaction form is generated and the link switch is executed. In this embodiment of the invention, when the switching trigger flag is valid, a switching transaction form is generated and the link switch is executed. It should be noted that the switching transaction form is used to solidify the triggering basis and contextual evidence, making the switch an auditable, verifiable, and rollback-capable transactional process.

[0119] Furthermore, during link switching, the default data plane exit can be switched from the service link to the target link, and link observation records can be generated according to step S1 after the switch for post-switch verification. It is easy to understand that the switching can be implemented through routing policy updates, primary / backup link switching, or primary selection via link aggregation; this invention does not limit the specific implementation method.

[0120] Finally, post-switch verification, failure rollback, cache update, and feedback constraints are performed. In this embodiment of the invention, a verification window is set after the switch, and the round-trip time and block error rate of the target link are summarized within the verification window to determine whether the switch has truly improved the service plane quality (for example, by calculating the average round-trip time and average block error rate of the target link within the verification window).

[0121] Furthermore, to clarify the success determination method, the difference between the baseline average before the handover and the average round-trip time and average block error rate calculated after the handover can be used to determine whether the improvement exceeds the corresponding threshold. Success is defined as when both satisfy the requirement of service quality improvement. It should be noted that the success determination simultaneously constrains latency and bit error rate, avoiding one-sided improvements such as improved latency but worsened bit error rate or vice versa, thus better meeting the stability requirements of remote control and real-time services. In practical applications, the improvement threshold can be configured according to the service type, but this invention does not limit the specific value.

[0122] When the successful handover flag does not meet the preset conditions, this embodiment performs a rollback to restore the service link and updates the risk count in the link fingerprint cache table to impose a penalty on the target link in subsequent scoring. It should be noted that the above update allows the verification results to be fed back to the risk penalty scoring step, significantly reducing the probability of repeated false handovers in scenarios with alternating strong reflections and occlusions. It is easy to understand that the cache table can record not only counts but also typical latency intervals or failure scenario labels, but this invention does not limit the way the cache fields are extended.

[0123] Reference Figure 2 , Figure 2 This is a schematic diagram of the CPE signal adaptive enhancement and link switching system according to an embodiment of the present invention.

[0124] like Figure 2 As shown, the CPE signal adaptive enhancement and link switching system proposed in this embodiment of the invention includes:

[0125] The acquisition module 10 is used to acquire the link identifier, reference signal received power, signal-to-interference-plus-noise ratio and block error rate of the serving link and the candidate link, and to generate the round-trip delay for the serving link to form a link observation record;

[0126] Extraction module 20 is used to form a short window sequence based on the link observation record, and extract the link shadow fingerprint from the short window sequence to characterize the quality fluctuation, block burst and delay spike characteristics of the current link in a short period of time.

[0127] The generation module 30 is used to perform change point detection to generate shadow event segment flags and shadow intensity by taking the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input.

[0128] Output module 40 is used to perform link state discrimination based on the link shadow fingerprint, the shadow event segment flag and the shadow intensity as observations, and output the probability of normal state, the probability of occlusion shadow state and the probability of multipath artifact state and the current state.

[0129] The control module 50 is used to perform signal enhancement control and link switching control based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block bit error rate of the candidate link.

[0130] Other embodiments or specific implementations of the CPE signal adaptive enhancement and link switching system of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0131] It is understood that in the description of this specification, references to terms such as "one embodiment," "another embodiment," "other embodiments," or "first embodiment to Nth embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0132] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0133] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for adaptive enhancement and link switching of CPE signals, characterized in that, The method includes the following steps: The system collects the link identifier, reference signal received power, signal-to-interference-plus-noise ratio, and block error rate of the serving link and candidate links, and generates round-trip delay for the serving link to form a link observation record. A short window sequence is formed based on the link observation records, and the link shadow fingerprint is extracted from the short window sequence to characterize the quality fluctuation, block bursts and delay spikes of the current link in a short period of time. Using the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input, change point detection is performed to generate shadow event segment markers and shadow intensity; specifically including: Based on the historical signal-to-interference-plus-noise ratio (SIR) sequence of the service links in the link observation records, a baseline mean SIR is generated, and the cumulative offset statistic of the SIR is updated based on the baseline mean SIR and the current SIR in the link shadow fingerprint. Based on the historical round-trip delay sequence of the service link in the link observation record, a mean round-trip delay baseline is generated, and based on the mean baseline and the current round-trip delay in the link shadow fingerprint, the cumulative offset statistic of the round-trip delay is updated. When the cumulative offset statistic of the signal-to-interference-plus-noise ratio or the cumulative offset statistic of the round-trip delay exceeds the corresponding threshold, the shadow event segment flag is determined to be valid, and the shadow intensity is generated based on the relative relationship between the cumulative offset statistic and the threshold. Specifically, based on the historical signal-to-interference-plus-noise ratio (SIR) sequence of the service links in the link observation records, the baseline mean SIR is generated and expressed as follows: ； In the formula, This represents the baseline mean of the signal-to-interference-plus-noise ratio (SINR). Indicates the historical length used to calculate the baseline; Indicates the sampling sequence number; Indicates the service link; This represents the signal-to-interference-plus-noise ratio (SIR) of historical samples; The cumulative offset statistic of the signal-to-interference-plus-noise ratio (SIR) is updated based on the baseline mean SIR and the current SIR in the link shadow fingerprint, and is expressed as follows: ； In the formula, This represents the cumulative offset statistic of the signal-to-interference-plus-noise ratio (SINR). Indicates the current sampling signal-to-interference-plus-noise ratio; This represents the drift compensation term; max(·) represents the maximum value operation; Specifically, based on the historical round-trip delay sequence of the service link in the link observation record, the mean of the round-trip delay baseline is generated and expressed as follows: ； In the formula, This represents the baseline mean of round-trip delay; Indicates the historical length used to calculate the baseline; Indicates the sampling sequence number; Indicates the service link; Indicates historical round-trip time delay; The cumulative offset statistic of the round-trip delay is updated based on the baseline mean and the current round-trip delay in the link shadow fingerprint, and is expressed as follows: ； In the formula, This represents the cumulative offset statistic for round-trip delay; Indicates the current round-trip time for sampling; Indicates the drift compensation term; The shadow intensity is generated based on the relative relationship between the cumulative offset statistic and the threshold, and is expressed as follows: ； In the formula, Indicates shadow intensity; and represents the CUSUM threshold for signal-to-interference-plus-noise ratio and round-trip delay, respectively; min(·) and max(·) represent the minimum and maximum value operations, respectively; Using the link shadow fingerprint, the shadow event segment flag, and the shadow intensity as observations, perform link state discrimination to output the probability of normal state, the probability of occlusion shadow state, the probability of multipath artifact state, and the current state; Based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block error rate of the candidate links, signal enhancement control and link switching control are performed.

2. The CPE signal adaptive enhancement and link switching method as described in claim 1, characterized in that, The steps for collecting link identifiers, reference signal received power, signal-to-interference-plus-noise ratio (SINNR), and block error rate of the serving link and candidate links, and generating round-trip time for the serving link to form a link observation record, specifically include: Obtain the link identifiers corresponding to the serving link and the candidate link, and read the corresponding reference signal received power, signal-to-interference-plus-noise ratio and block error rate respectively; Perform a lightweight probe on the control server to obtain the round-trip latency of the service link; The link identifier, reference signal received power, signal-to-interference-plus-noise ratio, block bit error rate, and round-trip time are written into the same link observation record at the same sampling time.

3. The CPE signal adaptive enhancement and link switching method as described in claim 1, characterized in that, Based on the link observation records, a short-window sequence is formed, and a link shadow fingerprint is extracted from the short-window sequence to characterize the quality fluctuations, block bursts, and delay spikes of the current link within a short period of time. Specifically, this includes: Extract the most recent consecutive records of a preset window length from the link observation records to form a short window sequence; Within the short window sequence, the signal-to-interference-plus-noise ratio fluctuation amplitude, the proportion of block error rate exceeding the preset error threshold, the maximum jump amplitude of adjacent samples of round-trip delay, and the average value of the reference signal received power of the statistical service link are obtained to obtain quality fluctuation characteristics for characterizing the degree of short-term quality degradation, error block burst characteristics for characterizing the degree of error block burst, delay spike characteristics for characterizing the degree of delay spike, and strength mean characteristics for characterizing the apparent signal strength level. The quality fluctuation feature, the block burst feature, the delay spike feature, and the intensity mean feature are combined to form a link shadow fingerprint.

4. The CPE signal adaptive enhancement and link switching method as described in claim 1, characterized in that, Using the link shadow fingerprint, the shadow event segment flag, and the shadow intensity as observations, a link state discrimination step is performed to output the probability of normal state, the probability of occlusion shadow state, the probability of multipath artifact state, and the current state. Specifically, this includes: The link state is divided into a set of states: normal state, occlusion shadow state, and multipath artifact state. An observation sequence is constructed based on the link shadow fingerprint, shadow event segment flag, and shadow intensity. The Hidden Markov Model is invoked to perform inference based on the observation sequence, and the state probability corresponding to each state is output. The state with the highest state probability is taken as the current state.

5. The CPE signal adaptive enhancement and link switching method as described in claim 1, characterized in that, Based on the current state, the probabilities of each state, and the shadow intensity, signal enhancement control is performed, specifically including: When the current state is an occlusion shadow state and the probability of the occlusion shadow state reaches the preset condition, the occlusion enhancement control mode is triggered, which increases the receiving gain target or increases the diversity merging priority based on the shadow intensity. When the current state is a multipath illusion state and the probability of the multipath illusion state reaches a preset condition, the illusion suppression control mode is triggered. If the triggering conditions for the occlusion enhancement control mode and the artifact suppression control mode are not met, the normal enhancement control mode is triggered.

6. The CPE signal adaptive enhancement and link switching method as described in claim 1, characterized in that, Based on the received power of the reference signal, signal-to-interference-plus-noise ratio (SINR), and block bit error rate of the candidate link, the link switching control steps are executed, specifically including: For each serving link and each candidate link, a candidate link score and a serving link score are generated based on the degree of improvement in signal-to-interference-plus-noise ratio, the degree of reduction in block bit error rate, and the level of received power of reference signal. The probability of occlusion and shadow states is used to increase the influence of signal-to-interference-plus-noise ratio on the candidate link score, and the probability of multipath artifact states is used to reduce the influence of reference signal received power on the candidate link score and increase the penalty of block error rate on the candidate link score. Based on the candidate link score and the service link score, read the historical failure count or risk count corresponding to the candidate link in the link fingerprint cache table, and use the failure count or risk count as a penalty factor for the link score to suppress repeated selection of high-risk links. Determine the candidate link with the largest link score after penalty as the target link, and calculate the score advantage between the target link and the service link. Based on shadow intensity and state probability, an adaptive switching threshold is generated and a link switching is triggered.

7. The CPE signal adaptive enhancement and link switching method as described in claim 6, characterized in that, Based on shadow intensity and state probability, an adaptive switching threshold is generated and a link switching step is triggered, specifically including: An adaptive handover threshold is generated based on shadow intensity, occlusion shadow state probability, and multipath artifact state probability to trigger link handover. The target link's score advantage is compared with the adaptive handover threshold to generate a candidate trigger flag when the score advantage exceeds the adaptive handover threshold. A continuity check is performed on the candidate trigger flags to generate a switching trigger flag when the candidate trigger flags continuously meet a preset number of times, and then a link switch is performed.

8. The CPE signal adaptive enhancement and link switching method as described in claim 7, characterized in that, The link switching procedure includes: When the switching trigger flag is valid, a switching transaction form is generated, and the trigger time, service link identifier, target link identifier, scoring advantage quantity, adaptive switching threshold, link shadow fingerprint, shadow intensity, current state and probability of each state are written into the switching transaction form; Based on the switching transaction form, the default data plane exit is switched from the service link to the target link, and after the switch, link observation records are continuously generated according to the preset verification window. The round-trip time and block error rate of the target link are summarized in the verification window and compared with the baseline before handover recorded in the handover transaction form, so as to generate a handover success flag when the round-trip time decreases and the block error rate decreases to a preset level. If the switch success flag does not meet the preset conditions, a rollback is executed to restore the service link, and the failure information is written to the link fingerprint cache table for the failure count or risk count corresponding to the target link.

9. A CPE signal adaptive enhancement and link switching system, used to execute the CPE signal adaptive enhancement and link switching method as described in any one of claims 1-8, characterized in that, include: The acquisition module is used to acquire the link identifier, reference signal received power, signal-to-interference-plus-noise ratio and block error rate of the serving link and candidate links, and to generate round-trip delay for the serving link to form a link observation record; The extraction module is used to form a short window sequence based on the link observation records, and extract the link shadow fingerprint from the short window sequence to characterize the quality fluctuation, block bursts and delay spikes of the current link in a short period of time. The generation module is used to perform change point detection to generate shadow event segment flags and shadow intensity by taking the link shadow fingerprint and the signal-to-interference-plus-noise ratio and round-trip delay in the link observation record as input; The output module is used to perform link state discrimination based on the link shadow fingerprint, the shadow event segment flag and the shadow intensity as observations, and output the probability of normal state, the probability of occlusion shadow state and the probability of multipath artifact state and the current state. The control module is used to perform signal enhancement control and link switching control based on the current state, the probability of each state, the shadow intensity, and the reference signal received power, signal-to-interference-plus-noise ratio, and block bit error rate of the candidate links.

Images