CPE cellular antenna switching method and system

Abstract

The invention relates to the technical field of antenna switching, in particular to a CPE cellular antenna switching method and system, and the method comprises the following steps: transmitting an in-band detection signal through a service antenna at a switching interval, and constructing a current service path feature vector according to a time delay spread spectrum fed back by an adjacent base station; after executing candidate antenna switching, immediately transmitting a homologous detection signal through a current antenna, and generating a target path feature vector; and calculating the similarity between the service path feature vector and the target path feature vector, maintaining switching when the similarity is lower than a preset judgment threshold, otherwise, switching back to the original service antenna and updating the path feature library. According to the method, the fine-grained difference of the propagation path structure is reflected, the accuracy and stability of path identification are effectively improved, and the method is particularly suitable for CPE application scenes under the multi-path rich or non-line-of-sight propagation condition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of antenna switching, in particular to a CPE cellular antenna switching method and system. BACKGROUND

[0002] With the popularization of 5G wireless access networks, cellular communication terminal devices (CPE, Customer Premises Equipment) are widely deployed in home, enterprise, remote site and other scenarios to replace traditional wired broadband and achieve fast and flexible broadband access capability. Modern CPE devices usually integrate multiple cellular communication antennas and support multiple-input multiple-output (MIMO) and multi-band collaborative communication technologies to cope with complex wireless channel environments and dynamic network conditions.

[0003] In actual deployment, due to the CPE device being in an indoor, window edge or semi-closed space with obstructions, the wireless signals received by the CPE device are often limited by factors such as multipath reflection, non-line-of-sight propagation, building obstruction and dynamic fading. In order to improve communication quality, the CPE device often needs to dynamically switch between multiple built-in or external antennas to select the antenna path with the best performance. However, the existing cellular antenna switching mechanism generally has the following problems: Insufficient accuracy of switching strategy based on intensity indicators: Traditional CPE devices mostly rely on a single indicator such as received signal strength (RSSI), reference signal received power (RSRP) or signal-to-interference-and-noise ratio (SINR) as the basis for switching, which is difficult to fully reflect the differences in propagation structure of different antenna paths in a multipath channel, and is prone to switching misjudgment or no performance improvement after switching.

[0004] Some existing solutions use real-time statistics or inserted probes on service data to make channel judgments, but often lack fine-grained resource control, which may conflict with normal service scheduling, affecting the continuity and stability of service transmission. Current mainstream CPE antenna switching methods mostly use static or preset switching thresholds, lack dynamic similarity discrimination mechanisms based on multi-dimensional path characteristics, and do not introduce feedback mechanisms for historical path information and multiple judgment results, making it difficult to cope with the time-varying nature and spatial heterogeneity of channel characteristics in complex environments. SUMMARY

[0005] The present application provides a CPE cellular antenna switching method and system, which introduces key mechanisms such as probe signal-driven path feature modeling, similarity calculation based on feature space, dynamic threshold decision and closed-loop feature library updating, to realize an interpretable and evolvable cellular antenna intelligent switching method.

[0006] A CPE cellular antenna switching method, comprising the following steps: S1. transmitting an in-band probe signal through the serving antenna in a switching interval, and constructing a current serving path feature vector according to a time delay spread spectrum fed back by a neighboring base station; S2. immediately transmitting a homologous probe signal through the current antenna after performing candidate antenna switching, and generating a target path feature vector; S3. calculating a similarity between the serving path feature vector and the target path feature vector, maintaining the switching when the similarity is lower than a preset judgment threshold, or switching back to the original serving antenna and updating a path feature library.

[0007] Optionally, in the switching interval of S1, a Zadoff-Chu sequence with autocorrelation characteristics is generated as the probe signal, and the probe signal is mapped to blank resource elements in a physical resource block used by the current serving antenna for transmission, and the probe signal only occupies the blank resource elements in the current physical resource block which are not modulated.

[0008] Optionally, in S1, a probe signal detection report fed back by the neighboring base station through an X2 interface is received, a power time delay distribution spectrum is extracted from the probe signal detection report, a main path relative intensity ratio, a time delay spread kurtosis and a multipath component entropy value are extracted and calculated based on the power time delay distribution spectrum, and finally combined into the serving path feature vector.

[0009] Optionally, the main path relative intensity ratio is calculated by a ratio of a path with the maximum received power to total received power of all paths; The time delay spread kurtosis is calculated by comparing a weighted average of fourth powers of time delay deviations of all paths with a time delay dispersion degree; The multipath component entropy value is obtained by calculating an entropy value of a power distribution after normalizing path powers.

[0010] Optionally, S2 includes, in a first subframe after performing candidate antenna switching, multiplexing the Zadoff-Chu sequence used in S1 as a homologous probe signal, and transmitting the homologous probe signal on blank resource elements of the same frequency point through the currently activated candidate antenna.

[0011] Optionally, the generation of the target path feature vector includes: Receiving a new power time delay distribution spectrum fed back by the same neighboring base station, extracting and calculating a main path relative intensity ratio, a time delay spread kurtosis and a multipath component entropy value, and combining the calculation results into a target path feature vector with the same dimension as S1.

[0012] Optionally, S2 further includes timestamp verification, specifically including recording a time difference Δt between a probe signal transmission time and a switching completion time, and discarding the target path feature vector when Δt>1ms.

[0013] Optionally, S3 specifically includes: S31, project the service path feature vector and the target path feature vector into a three-dimensional orthogonal feature space respectively, the three-dimensional orthogonal feature space is composed of three physical feature dimensions of main path relative intensity ratio, time delay spread kurtosis and multipath component entropy value, and is used for comprehensively representing the propagation characteristics of the path; S32, in the three-dimensional orthogonal feature space, calculate the difference of the service path feature vector and the target path feature vector in three dimensions, and introduce an adaptive weight coefficient determined by the service path feature, weight process the difference values of different dimensions, and then calculate the comprehensive difference degree between the paths as the evaluation index of the similarity of the two. S32, according to the size of the channel coherence bandwidth, generate a decision threshold, and compare the calculated path difference value with the decision threshold; if the difference value is higher than the decision threshold, it means that the target path and the current service path are obviously different, and the current switching is maintained; if the difference value is lower than the decision threshold, it is considered that the switching is invalid, and the back switching operation is triggered.

[0014] Optionally, the S3 further includes path feature library updating, specifically including: in the case of back switching, accumulating and increasing the confidence of the service path feature vector; when the accumulation reaches a preset threshold, proportionally fusing the current service path feature vector and the target path feature vector, constructing a new path representation, and using it to replace the original feature vector.

[0015] A switching system of a CPE cellular antenna is used to realize the switching method of the CPE cellular antenna, and includes the following modules: A detection signal emitting module is used to emit in-band detection signals through the service antenna in a switching interval, and emit homologous detection signals through the candidate antenna after switching; A feature vector generating module is used to construct the service path feature vector and the target path feature vector according to the time delay spread spectrum fed back by the adjacent base station respectively; A similarity calculating module is used to calculate the similarity between the service path feature vector and the target path feature vector; A decision and control module is used to judge whether to maintain the current switching or back switch to the original service antenna according to a preset decision threshold, and update the path feature library when back switching.

[0016] The beneficial effects of the present application are: The application inserts a Zadoff-Chu (ZC) sequence probe signal with ideal autocorrelation characteristics into an idle time slot of service transmission, and transmits the signal in a blank resource element (RE) of a current serving antenna, so that the power delay profile of the service path can be extracted with high precision without interrupting service communication; and a highly distinguishable service path feature vector is constructed by modeling a combination of multi-dimensional characteristics such as a main path relative strength ratio, a delay spread kurtosis, and a multi-path component entropy value. Compared with a conventional switching strategy that relies on a single signal strength indicator such as RSSI or SINR, the scheme can reflect fine-grained differences in the propagation path structure, effectively improve the accuracy and stability of path identification, and is particularly suitable for CPE application scenarios under rich multi-path or non-line-of-sight propagation conditions.

[0017] The application actively multiplexes a homologous ZC sequence in the first subframe after antenna switching, re-emits a probe signal on the same frequency domain resource by the current candidate antenna, acquires a target path feature vector, and calculates the similarity between the target path feature vector and the service path by using a three-dimensional weighted Euclidean distance, while introducing a dynamic weight adjustment mechanism driven by the characteristics of the service path and a threshold adjustment mechanism based on coherent bandwidth, so that a highly adaptive path matching decision strategy is realized. The mechanism not only effectively avoids mismatching caused by frequency domain differences, time delays, or base station feedback deviations, but also automatically adjusts the decision criteria according to the complexity of the environment, improves the adaptability and robustness of the switching strategy in dense urban areas, suburban areas, and high dynamic scenarios, and reduces the frequency of mis-switching and back-switching.

[0018] The application designs a path feature library closed-loop update mechanism based on confidence accumulation and vector fusion according to the path stability information reflected by multiple back-switching: after detecting a back-switching event, the confidence of the service path is automatically increased, and the service path and the target path are proportionally fused to generate a new vector; when the confidence accumulates to a preset threshold, the original feature vector is automatically replaced to construct a path model that is more consistent with the actual current channel. Not only the memory and evolution of path identification are enhanced, but also new observation information can be effectively absorbed to dynamically correct the path discrimination result, so that the adaptability and identification stability of the feature model are improved in scenarios of frequent switching or gradual channel drift. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only illustrate the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0020] Fig. 1 The method flowchart of the embodiment of the application is shown in the figure. Fig. 2The schematic diagram of the probe signal of the embodiment of the present application. DETAILED DESCRIPTION

[0021] The present application is described in detail below with reference to the drawings and specific embodiments, and other alternative ways can also be implemented by those skilled in the art; and the drawings are only used to more specifically describe the embodiments, and are not intended to specifically limit the present application.

[0022] As shown in the figure, a switching method of a CPE cellular antenna comprises the following steps: Figs. 1-2 S1. In the switching interval, a probe signal is transmitted by the serving antenna, and a current serving path feature vector is constructed according to the time delay spread spectrum fed back by the adjacent base station; S2. After the candidate antenna switching is performed, a homologous probe signal is immediately transmitted by the current antenna to generate a target path feature vector; S3. The similarity of the serving path feature vector and the target path feature vector is calculated, and when the similarity is lower than a preset judgment threshold, the switching is maintained, otherwise the original serving antenna is switched back and the path feature library is updated.

[0023] In S1, a Zadoff-Chu (ZC) sequence with autocorrelation characteristics is generated as a probe signal in the idle time slot of the service transmission, i.e. in the switching interval, and the probe signal is mapped to the blank resource elements (REs) in the physical resource block (PRB) used by the current serving antenna for transmission.

[0024] The ZC sequence has ideal autocorrelation characteristics, with sharp main peak and rapid sidelobe decay, which can improve the time delay resolution of the multipath path and is suitable for time domain detection.

[0025] Blank RE utilization principle: the probe signal only occupies the blank resource elements (REs) in the current PRB that are not modulated, so as to avoid interfering with the transmission of the service data.

[0026] The transmission power of the probe signal is 6-10 dB lower than the service signal power, which ensures that it will not interfere with the communication of the adjacent cell.

[0027] The probe signal detection report fed back by the adjacent base station through the X2 interface is received, and the power time delay distribution spectrum is extracted therefrom. The power time delay distribution spectrum is the received power size distribution of a certain probe signal received by the base station at different time delays (times), and the parameters included in the distribution spectrum are the time delay value of each path, the received power of each path and the total number of paths, which are used as the basic data source to calculate the following three feature parameters: 1. Main path relative intensity ratio: ; wherein, is the received power of the maximum power path,​ total received power of all multipath components; 2. kurtosis of delay spread: ; where, denotes path delay, denotes mean delay (average delay), denotes delay standard deviation, denotes expectation operator, which is calculated as follows: 2.1. : extract delay values of all multipath components from power-delay profile and their corresponding normalized power weights , which satisfy , can be used as probability weights.

[0028] 2.2. calculate mean delay : sum all delay values multiplied by their corresponding power weights to get weighted average, which is mean delay.

[0029] 2.3. perform transformation for each component: for each delay , calculate its deviation from mean , and take fourth power, i.e. .

[0030] 2.4. perform weighted summation: multiply each fourth power result by its corresponding power weight , and sum all products, which is to complete expectation calculation.

[0031] 2.5. result output: the above weighted summation is the result of expectation operation , which is used for subsequent kurtosis calculation.

[0032] 3. entropy of multipath components: ; where, is total number of multipath components, is normalized received power of th multipath component, i.e. , denotes original received power of th multipath component, denotes total received power of all multipath components.

[0033] Finally, the above three features are combined into service path feature vector: .

[0034] The method for transmitting ZC probe signal in idle time slot is as follows: a) : Idle time slot positioning: In the service transmission scheduling period, monitor the state of the resource elements (RE) contained in the current physical resource block (PRB); identify the blank RE set not scheduled for data channel or control channel bearing from it; mark these blank REs as resource sites where the probe signal can be inserted.

[0035] b) : Probe signal sequence generation: select a predefined root index , generate a Zadoff-Chu (ZC) sequence with a length of according to the standard formula: ; wherein, represents the th Zadoff-Chu (ZC) sequence element, is the imaginary unit, satisfying , is the constant of the circle, represents the root index of the ZC sequence (positive integer, satisfying prime with ), represents the sequence index variable, represents the length of the sequence (positive integer), in order to prevent spectral leakage, the sequence is processed by Hamming window and normalized, if the total number of blank REs is more than the length of the ZC sequence, it can be repeated in the time-frequency domain; if it is less than the length of the ZC sequence, it is truncated by equidistant sampling.

[0036] c) : Signal insertion and mapping: map the ZC sequence to the blank RE set in order to ensure that its time-frequency position does not overlap with the service data RE, if the blank RE distribution in the PRB is discontinuous, frequency hopping or cross-PRB insertion strategy can be used to avoid disturbing the resource scheduler.

[0037] d) : Transmit power control: set the independent transmission power coefficient of the ZC probe signal α ∈ [−10,−6] dB, dynamically adjust according to the current CPE transmission power budget and frequency domain resource utilization; the transmission power control module needs to ensure that the probe signal is below the noise in the adjacent cell or other terminal sensing, to avoid false triggering of interference processing.

[0038] e) : Transmission execution and time control: the insertion time of the probe signal is strictly limited within the switching interval period, and the duration does not exceed 1ms, after transmission, immediately start receiving the probe report returned by the adjacent base station, and enter the feature parameter extraction stage.

[0039] The execution of candidate antenna switching refers to switching the service antenna originally connected by the current CPE device to a preset candidate antenna, so that the receiving and transmitting channel of the wireless signal is borne by the candidate antenna. The CPE device is internally integrated with multiple cellular antennas, and the "switching" operation refers to controlling the switch matrix or antenna control module of the radio frequency front end to switch the receiving and transmitting link from the original service antenna to the target candidate antenna. The switching action is completed within a scheduling period or a switching interval, and does not affect the upper layer protocol stack. After switching, detection is immediately performed without a complete connection reconstruction process. By switching to other antennas to obtain another set of propagation paths, it is detected whether there is a more optimal channel environment under the antenna. It is a local path switching behavior rather than a complete cell or base station switching. The "execution of candidate antenna switching" is to replace the currently used cellular communication antenna with another selectable antenna in order to evaluate the new path.

[0040] In S2, in the first subframe after the completion of the candidate antenna switching, the following operations are performed to actively evaluate the new antenna path characteristics: S21, probe signal transmission: sequence replication is performed, the Zadoff-Chu (ZC) sequence used in S1 is directly multiplexed, the same root index r and cyclic shift value are maintained, the homogeneity of the transmitted signal and the original service path probe signal is ensured, the same physical resource block (PRB) and resource element (RE) position as S1 is used, the corresponding blank RE is transmitted on the currently activated candidate antenna to avoid interfering with the transmission of service data, thereby completing the time-frequency synchronization, introducing a time constraint, and ensuring that the current channel state has not changed significantly. The probe signal must be transmitted within the first subframe after the completion of the switching.

[0041] Multiplexing the ZC sequence of S1 can maintain signal homogeneity. Multiplexing the same Zadoff-Chu (ZC) sequence (consistent root index and cyclic shift value) can ensure the correlation characteristics of the probe signal, including the peak position and sidelobe structure, which remain consistent under different antennas. If S2 uses a different sequence from S1, a new waveform effect will be introduced, which may cause differences in path characteristics from both the antenna spatial characteristics and the sequence itself, resulting in inaccurate feature comparison. The main purpose is to ensure that the target path feature vector and the service path feature vector are only affected by the difference in spatial propagation path when constructing, thereby ensuring the consistency and reliability of path matching judgment.

[0042] The same PRB and RE are used in time-frequency synchronization, the response of a wireless channel is highly time-frequency correlated, the transmission resource (PRB, RE) is kept consistent with S1, the feature disturbance caused by factors such as frequency offset and subcarrier mismatch can be avoided, the idle RE is selected for transmission, the scheduling resource is not occupied, and the upper-layer service is not interfered, and the applicability of the scheme in a real-time service concurrent environment is ensured. In the application, the two detections are compared under the same spectrum position and resource mapping structure, so that the path feature comparison result truly reflects the physical difference of different antenna receiving paths, rather than the pseudo difference caused by scheduling or frequency domain difference.

[0043] In the LTE / 5G system, the wireless channel between the user terminal and the base station fluctuates over time. For a static or low-speed scenario, the channel coherence time is about 10-50 ms, and in a high-speed scenario, the coherence time can be less than 5 ms. If the detection operation is delayed for more than 1 ms, the original service path and the target path can no longer be in the same channel state, resulting in invalid feature vector comparison. Therefore, a time constraint (must be completed within 1 ms) is set. The detection operation is ensured to occur in the shortest time after the antenna switching is completed, the channel state is prevented from changing due to time evolution, the authenticity and timeliness of the path feature comparison are ensured, and the accuracy and reliability of the switching decision are improved.

[0044] S22, target path feature vector generation: The detection report of the probe signal returned by the adjacent base station through the X2 interface is received, and the updated power delay profile is extracted.

[0045] The following three feature parameters are extracted using the same calculation as S1: main path relative intensity ratio, delay spread kurtosis, and multipath component entropy value.

[0046] S23, feature vector construction: the three parameters are combined into a three-dimensional target path feature vector: ; wherein , , respectively represent the three feature parameters calculated under the current candidate antenna path. The time difference between the probe signal transmission time and the antenna switching completion time is recorded: ; if ms, the target path feature vector is discarded, and the invalid detection is regarded as invalid, so as to avoid evaluation distortion caused by channel state change, represents the time difference between the probe signal transmission time and the antenna switching completion time, which is used to judge whether the channel state is still stable, represents the time stamp of the actual transmission of the probe signal, which is recorded by the CPE through a hardware timer at the transmission moment, Timestamp indicating the time when the candidate antenna completes the handover, i.e. the time point when the physical radio path switching action is completed.

[0047] S3 specifically comprises: S31, feature space mapping: projecting the service path feature vector and the target path feature vector into a three-dimensional orthogonal feature space, where each coordinate axis is defined as follows: X-axis: relative strength ratio of the main path (indicating the degree of dominance of the direct path); Y-axis: kurtosis of the delay spread (indicating the steepness of the multipath distribution); Z-axis: entropy value of the multipath component (indicating the dispersion degree of the multipath energy).

[0048] S32, dynamic similarity calculation: calculating the distance in the three-dimensional feature space, indicating the difference between the two path feature vectors: ; wherein: represents the difference between the service and target paths on the X-axis; represents the difference between the service and target paths on the Y-axis; represents the difference between the service and target paths on the Z-axis; represents the Y-axis weight, which increases with the increase of the service path kurtosis, represents the kurtosis of the delay spread in the service path feature vector, describing the steepness of the multipath delay distribution (the larger the value, the sharper the distribution); represents the Z-axis weight, which decreases with the increase of the service path entropy value, represents the entropy value of the multipath component in the service path feature vector, describing the dispersion degree of the multipath power among the paths (the larger the value, the more dispersed the energy).

[0049] The three-dimensional Euclidean distance is used to calculate the difference between the service path and the target path in the "path feature space", which is a classic vector similarity measurement method; each axis represents a physically significant channel feature (main path dominance, delay distribution steepness, multipath energy distribution), and the smaller the three-dimensional distance, the more similar the two paths are in physical characteristics; this method is sensitive to dimensional changes, intuitive and visual, and has low computational cost, making it suitable for deployment on embedded CPE devices. Each coordinate axis corresponds to a channel feature parameter with physical meaning, which can reflect the different propagation mechanisms of the antenna receiving path: X-axis: degree of dominance of the direct component; Y-axis: state of multipath aggregation or dispersion; Z-axis: whether the energy distribution is concentrated; This design makes the distance metric not an abstract vector difference, but an interpretable propagation behavior difference metric.

[0050] The contribution of different feature dimensions to the "path differentiation ability" may be different in different scenarios, so dynamic weights are introduced to control the adaptive weight coefficient as follows: Y-axis weight : The steeper the service path, the more sensitive the Y-axis (kurtosis) change to channel changes; Z-axis weight : The more dispersed the service path energy, the greater the entropy value, and the lower the differentiation ability of the Z-axis change, so the weight is reduced.

[0051] Adaptive feature weighting mechanism is embodied, and the feature of the service path itself "reverses" which dimension should be more important in this evaluation.

[0052] Traditional antenna switching is mainly based on single strength parameters such as RSSI / RSRP / SINR, which is easily disturbed by short-term fluctuations. The present application improves the robustness of switching judgment by introducing multi-dimensional feature vector space difference measurement + weight adjustment + dynamic threshold mechanism. The weight mechanism is driven by the current state of the service path, and can adapt to the influence of different environmental characteristics on channel changes: In the shielding environment, the kurtosis difference is more sensitive -> the Y-axis weight becomes larger; In the scattering environment, the entropy value tends to be high but the non-deterministic is strong -> the Z-axis weight becomes smaller; This mechanism ensures that the path matching model has high portability in multiple scenarios, strong multipath, and non-static environments.

[0053] S33, threshold adaptive decision: set decision threshold , which is related to the frequency selectivity characteristics of the channel: ; wherein, is the basic threshold (0.65 by default in dense urban areas, and 0.55 in suburban areas), represents the channel coherence bandwidth, with a unit of MHz, reflecting the frequency selectivity degree of the channel (not more than 50MHz). The higher the T value: allows the path characteristics to be more similar and switch, high sensitivity, high risk of mis-cut; The lower the T value: only when the path characteristics are significantly different, switching is allowed, low sensitivity, more robust. The decision rule is as follows: If , it is considered that the target path and the service path are significantly different, and the current antenna switching is maintained; If , it is considered that the switching path has no significant advantage, and the back-cut operation is performed.

[0054] Dense urban area: the building is dense, the multipath reflection is rich, the channel changes rapidly, it is difficult to be stable for a long time, the path characteristic fluctuates frequently, a higher threshold (0.65) needs to be set to avoid excessive sensitivity leading to frequent switching; 1Rural / empty scene: sparse multipath, direct dominant, relatively stable channel, the path difference received by different antennas is more real and reliable, a lower threshold (0.55) should be set to improve the discrimination accuracy and prevent misidentification as'similar path' and miss effective switching.

[0055] By setting different basic thresholds T0 for different scenes, and cooperating with the dynamic adjustment function, the decision mechanism has stronger adaptability to urban dense areas, suburbs, indoor scenes and the like. In the urban area where the channel changes dramatically, the switching threshold is appropriately relaxed, which helps to avoid false switching caused by short-term small fluctuations; and in the suburbs where the channel is stable, the threshold is tightened, which helps to improve the sensitivity of identifying'really different paths'.

[0056] S34, path feature library update: when the decision is switched back, the following update operation is performed to enhance the adaptive ability of path judgment: S341. Confidence accumulation: the service path feature vector in the path feature library is increased by a constant , indicating that the stability of the path is confirmed again.

[0057] In the present application, represents the confidence enhancement value of the service path feature after each'switch back' event, which is used to construct a dynamically updated path feature library. When a path is repeatedly confirmed to be 'better' after multiple switching failures, the cumulative confidence reaches the update threshold, triggering vector fusion and replacement. The confidence threshold for triggering vector replacement in the present application is 0.6 (i.e. 3 consecutive switch-back confirmations are required), and , 3 confirmations can just reach the update condition, which matches the common robustness judgment standard of 'three confirmations in agreement'.

[0058] S342. Vector fusion update: construct a new path feature vector after fusion: ; represents retaining 70% of the original service path features. And introduce 30% of the new information of the current target path to update the path gradually.

[0059] S343. Update trigger condition: when the cumulative confidence increases (i.e. 3 consecutive switch-backs), replace the original path feature vector with .

[0060] A switching system of a CPE cellular antenna is used to implement the switching method described above, which comprises the following modules: The probe signal emitting module is configured to emit an in-band probe signal through the serving antenna during a switching interval and emit a homologous probe signal through the candidate antenna after switching; The feature vector generating module is configured to construct a serving path feature vector and a target path feature vector respectively according to time delay spread spectrums fed back by adjacent base stations. The similarity calculating module is configured to calculate a similarity between the serving path feature vector and the target path feature vector. The decision and control module is configured to determine whether to maintain the current switching or switch back to the original serving antenna according to a preset decision threshold, and update the path feature library when switching back.

[0061] The present application encompasses any alternatives, modifications, equivalent methods and schemes made on the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details by those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0062] The above is only the preferred embodiment of the present application, and it should be pointed out that for those skilled in the art, without departing from the principle of the present application, a number of improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A switching method of a CPE cellular antenna, characterized by, The method comprises the following steps: S1. transmitting an in-band probe signal through a serving antenna during a switching interval, constructing a current serving path feature vector according to a time delay spread spectrum fed back by a neighboring base station; S2. immediately transmitting a homologous probe signal through the current antenna after performing candidate antenna switching, generating a target path feature vector; S3. calculating the similarity of the serving path feature vector and the target path feature vector, maintaining switching when the similarity is lower than a preset judgment threshold, otherwise, switching back to the original serving antenna and updating the path feature library.

2. The method of claim 1, wherein the CPE cellular antenna switching method is characterized by, In the switching interval of S1, a Zadoff-Chu sequence with autocorrelation characteristics is generated as the probe signal, and the probe signal is mapped to blank resource elements in a physical resource block used by the current serving antenna for transmission, and the probe signal only occupies the unmodulated blank resource elements in the current physical resource block.

3. The method of claim 1, wherein the CPE cellular antenna switching method is characterized by, In S1, a probe signal detection report fed back by the neighboring base station through an X2 interface is received, a power time delay distribution spectrum is extracted from the probe signal detection report, a main path relative intensity ratio, a time delay spread kurtosis and a multipath component entropy value are extracted and calculated based on the power time delay distribution spectrum, and finally combined into the serving path feature vector.

4. The method of claim 3, wherein the CPE cellular antenna switching method is characterized by, The main path relative intensity ratio is calculated by the ratio of the path with the maximum received power to the total received power of all paths; The time delay spread kurtosis is calculated by comparing the weighted average of the fourth power of the time delay deviation of all paths with the time delay dispersion degree; The multipath component entropy value is obtained by calculating the entropy value of the power distribution after normalizing the power of each path.

5. The method of claim 1, wherein, S2 includes the first subframe after performing candidate antenna switching, multiplexing the Zadoff-Chu sequence used in S1 as a homologous probe signal, and transmitting it through the currently activated candidate antenna on the blank resource elements of the same frequency.

6. The method of claim 5, wherein, The generation of the target path feature vector comprises: Receiving a new power time delay distribution spectrum fed back by the same neighboring base station, extracting and calculating the main path relative intensity ratio, the time delay spread kurtosis and the multipath component entropy value, and combining the calculation results into a target path feature vector with the same dimension as S1.

7. The method of claim 6, wherein, S2 also includes timestamp verification, specifically including recording the time difference Δt between the probe signal transmission time and the switching completion time, and discarding the target path feature vector when Δt>1ms.

8. The method of claim 1, wherein, S3 specifically comprises: S31, projecting the serving path feature vector and the target path feature vector into a three-dimensional orthogonal feature space, respectively, the three-dimensional orthogonal feature space is composed of three physical feature dimensions of the main path relative intensity ratio, the time delay spread kurtosis and the multipath component entropy value, and is used to fully characterize the propagation characteristics of the path; S32, in the three-dimensional orthogonal feature space, the difference values of the serving path feature vector and the target path feature vector in three dimensions are calculated, and an adaptive weight coefficient determined by the serving path feature is introduced to weight the difference values in different dimensions, and then the comprehensive difference degree between the paths is calculated as an evaluation index of the similarity of the two. S32, generating a decision threshold according to the size of the channel coherence bandwidth, and comparing the calculated path difference value with the decision threshold; if the difference value is higher than the decision threshold, it means that the target path and the current service path are significantly different, and the current switching is maintained; if the difference value is lower than the decision threshold, it is considered that the switching is invalid, and a back switching operation is triggered.

9. The method of claim 8, wherein, The S3 also includes path feature library updating, specifically including: in the case of back switching, the confidence of the service path feature vector is accumulated and increased; when the accumulation reaches a preset threshold, the current service path feature vector and the target path feature vector are fused in proportion to construct a new path representation, and are used to replace the original feature vector.

10. A switching system of a CPE cellular antenna for implementing the switching method of the CPE cellular antenna according to any one of claims 1-9, characterized in that, It includes the following modules: A probe signal emitting module for emitting in-band probe signals through the service antenna during the switching interval, and emitting homologous probe signals through the candidate antenna after switching; A feature vector generating module for constructing a service path feature vector and a target path feature vector according to the delay spread spectrum fed back by adjacent base stations respectively; A similarity calculating module for calculating the similarity between the service path feature vector and the target path feature vector; A decision and control module for judging whether to maintain the current switching or back switch to the original service antenna according to a preset decision threshold, and updating the path feature library when back switching.

Images