A multi-device synchronous communication method of a WiFi7 chip module

Abstract

The application discloses a kind of multi-device synchronous communication methods of WiFi7 chip module, it is related to the technical field of wireless communication, the method comprises: according to the symbol stream corresponding to original link in physical layer issued by the radio frequency front end of WiFi7 chip module determines effective received signal energy;Statistical transmission layer of WiFi7 chip module in all the TCP acknowledgement message corresponding to original link in preset time window the total number of application layer effective data bytes confirmed, obtain effective data volume;Request the total amount of data sent and the physical layer transmission energy in the same time window to target device;According to the ratio of effective received signal energy and physical layer transmission energy, and the ratio of effective data volume and total amount of data sent, determine the entropy capacity corresponding to each time window;If the entropy capacity corresponding to continuous preset number of time windows is less than preset first entropy capacity threshold, then determine original link as invalid link.The application improves the aggregation performance of MLO when multiple communication devices access.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communications, and in particular to a multi-device synchronous communication method for a WiFi7 chip module. Background Art

[0002] Multi-Link Operation (MLO) is one of the core enhancements in the Wi-Fi 7 (IEEE 802.11be) standard. It allows a Wi-Fi device (such as an access point (AP) or client (STA)) to simultaneously establish and maintain multiple independent physical links across multiple frequency bands (e.g., 2.4 GHz, 5 GHz, and 6 GHz) or channels (within the same frequency band). These links can be aggregated to provide higher overall throughput, lower latency, and improved reliability for single or multiple smart devices (such as mobile phones, tablets, and smart home appliances). Ideally, MLO intelligently distributes data flows across different links for transmission, and performs switching or load balancing when link quality degrades.

[0003] However, in complex real-world wireless environments (such as those with thick walls, metal obstacles, co-channel / adjacent-channel interference, and interference from other wireless devices), MLO faces a key challenge: performance degradation of a single link can seriously hinder the performance of the entire aggregated link.

[0004] A common existing technology is link switching based on bit error rate (BER). The basic principle is to continuously monitor the BER of each link (the ratio of the number of bits decoded incorrectly by the receiver to the total number of bits transmitted). When the BER of a link exceeds a preset threshold, the system deems the link quality poor and triggers a switching operation. For example, the data flow can be switched from the currently heavily interfered 6GHz band to the 2.4GHz band, which has better penetration but lower data rates.

[0005] However, the BER detection method has a fatal technical flaw: it can only detect link problems after bit errors have already occurred at the physical layer (PHY layer). It cannot identify pseudo-active link conditions: that is, when the physical layer is active and normal, but the application layer is inoperative. In this pseudo-active link state, the BER detection method mistakenly assumes the link is healthy and does not trigger a handover. To make matters worse, the MLO mechanism in the WiFi7 chip will still attempt to aggregate data streams onto this failed link. This leads to even more serious side effects. For example, the protocol overhead of the MLO aggregation operation itself wastes bandwidth, causing a significant drop in effective bandwidth.

[0006] Therefore, how to improve the MLO aggregation performance when multiple communication devices access MLO has become a technical problem that needs to be solved urgently. Summary of the Invention

[0007] The technical problem solved by the present invention is that when multiple communication devices are connected to the WiFi7 chip module, there is a situation where the MLO aggregation performance is low.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: a multi-device synchronous communication method for a WiFi7 chip module, the method being applied to a communication device equipped with a WiFi7 chip module, wherein the communication device transmits signals to a target device among multiple connected smart devices via an original link, the method comprising:

[0009] Determine the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the radio frequency front end of the WiFi7 chip module;

[0010] Counting the total number of valid data bytes of the application layer confirmed by all TCP confirmation messages corresponding to the original link at the transport layer of the WiFi7 chip module within a preset time window to obtain the valid data volume;

[0011] Requesting the target device for a total amount of data to be sent and a physical layer transmission energy within the same time window;

[0012] Determining the entropy capacity corresponding to each time window according to the ratio of the effective received signal energy to the physical layer transmit energy, and the ratio of the effective data volume to the total amount of transmitted data;

[0013] If the entropy capacity corresponding to a preset number of consecutive time windows is less than a preset first entropy capacity threshold, the original link is determined to be a failed link.

[0014] Preferably, before determining the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the RF front end of the WiFi7 chip module, the method further includes:

[0015] In response to the multi-link aggregation symbol stream of the physical layer issued by the radio frequency front end of the WiFi7 chip module;

[0016] A subcarrier symbol sequence corresponding to the original link is separated from the multi-link aggregated symbol stream to obtain a symbol stream corresponding to the original link.

[0017] Preferably, determining the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the radio frequency front end of the WiFi7 chip module includes:

[0018] Extracting a direct wave component from the subcarrier symbol sequence by a preset minimum mean square error channel equalizer;

[0019] The effective received signal energy is determined according to the direct wave component.

[0020] Preferably, before determining the effective received signal energy according to the direct wave component, the method further includes:

[0021] determining, according to the direct wave component, whether the subcarrier symbol sequence experiences wall attenuation during transmission;

[0022] If yes, determining the signal transmission time according to the timestamp of the direct wave component and the reception time of the direct wave;

[0023] Determining a transmission distance based on the signal transmission time and a preset transmission speed;

[0024] The energy value of the direct wave component is corrected according to the preset wall material attenuation coefficient and the transmission distance.

[0025] Preferably, the TCP confirmation message includes five-tuple data; before counting the total number of application layer valid data bytes confirmed by all TCP confirmation messages corresponding to the original link at the transport layer of the WiFi7 chip module within a preset time window to obtain the valid data amount, the method further includes:

[0026] Determine the link to which the TCP confirmation message belongs through the five-tuple data of the TCP confirmation message of the transport layer;

[0027] The TCP acknowledgment message whose link is the original link is determined as the TCP acknowledgment message corresponding to the original link.

[0028] Preferably, after determining the original link as a failed link if the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, the method further includes:

[0029] Generate a link failure signal, wherein the link failure signal includes a link identifier and a MAC close operation code of the original link;

[0030] Shielding the resource allocation request of the original link at the local MAC layer of the WiFi7 chip module, and selecting a target link from multiple communication links of the WiFi7 chip module, where the target link is different from the original link;

[0031] The link failure signal is sent to the target device via the target link, so that the target device closes the MAC layer resource allocation of the original link according to the link identifier in response to the MAC close operation code.

[0032] Preferably, after sending the link failure signal to the target device through the target link so that the target device closes the MAC layer resource allocation of the original link according to the link identifier in response to the MAC close operation code, the method further includes:

[0033] Sending a test data stream to the target device via the original link at a preset time interval;

[0034] determining an entropy capacity corresponding to a test data stream;

[0035] If the entropy capacity corresponding to the test data flow is greater than a preset second entropy capacity threshold, the resource allocation authority of the original link is restored, wherein the second entropy capacity threshold is less than the first entropy capacity threshold.

[0036] Preferably, the test data stream includes a UDP probe packet, and the UDP probe packet includes a forced response identifier, so that the target device returns a TCP confirmation message after detecting the forced response identifier.

[0037] Preferably, before determining the original link as a failed link if the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, the method further includes:

[0038] If the entropy capacity corresponding to a certain time window is less than a preset third entropy capacity threshold, the original link is determined as a failed link, wherein the third entropy capacity threshold is less than the second entropy capacity threshold.

[0039] Preferably, if the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, determining the original link as a failed link further includes:

[0040] Determine whether a network congestion flag appears at the transport layer according to an ECN field in a header of an IP message associated with the transport layer;

[0041] If there is no network congestion at the transport layer and the entropy capacity corresponding to a preset number of consecutive time windows is less than a preset first entropy capacity threshold, the original link is determined to be a failed link.

[0042] The beneficial effects of the present invention are as follows: the throughput efficiency ratio of the transmission layer is characterized by the ratio of the effective received signal energy to the physical layer transmission energy, and the energy efficiency ratio of the physical layer is characterized by the ratio of the effective data volume to the total amount of transmitted data, so that the obtained entropy capacity can integrate the data of the physical layer and the transmission layer, and truly reflect the communication capability of the original link, so as to make a more accurate judgment on whether the original link is invalid, so that the WIFI7 module can abandon the aggregation of the failed original link and improve the aggregation performance of MLO. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the basic flow of a multi-device synchronous communication method for a WiFi7 chip module provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0045] The inventors of this application discovered in actual use that BER detection methods cannot identify pseudo-active links. For example, in the 6GHz band, devices can complete normal protocol handshakes (e.g., sending RTS / CTS frames and receiving ACK frames). From the PHY layer's perspective, the signal is received, and the decoder may not report a large number of errors (the BER may be low or even normal), making the link appear healthy. The presence of ACK feedback can mislead the BER detection method into believing the link is healthy. However, due to severe signal attenuation (e.g., passing through multiple load-bearing walls), persistent strong interference, or other reasons, the transmission of valid data has actually failed, indicating a failure at the transport or application layer. For example, if data packets are severely lost during transmission (even if the PHY layer receives the signal, the decoding fails the upper layer check, or the signal is too weak to be correctly demodulated), the sender will repeatedly retransmit the lost data packets. When the number of retransmissions exceeds the maximum number specified by the TCP protocol (usually controlled by the tcp_retries2 sysctl parameter), the TCP send window closes, and the effective throughput of the TCP connection drops to zero. At this point, for the application layer (such as video streaming and file transfer), this link is actually invalid and cannot transmit any valid data.

[0046] In this pseudo-active state, the BER detection method mistakenly deems the link healthy and does not trigger a handover. To make matters worse, the Wi-Fi 7 chip's Multi-Link Lockout (MLO) mechanism still attempts to aggregate data streams onto the failed link. MLO's aggregation operation itself requires additional protocol overhead (such as multi-link management frames and more complex scheduling signaling). When the system continuously attempts to send data onto a link that is unable to transmit valid data, precious radio channel resources and time are used to transmit retransmitted packets or management frames that are ultimately discarded, rather than payloads. Because resources are occupied by the inactive link and the overhead of the aggregation mechanism itself increases, the total effective throughput of the entire MLO aggregated link (i.e., the actual available bandwidth perceived by the application layer) is actually lower than when the failed link is omitted and only the remaining healthy links are aggregated, resulting in a pseudo-rate trap.

[0047] Therefore, the inventors of the present application found that there is an urgent need for a method that can accurately identify such pseudo-active links when the physical layer signal appears normal but the application layer transmission has actually failed, and allow the MLO mechanism to actively give up aggregating the failed link, thereby improving the overall aggregation performance.

[0048] Based on this, this application proposes a multi-device synchronous communication method for WiFi7 chip modules, which introduces entropy capacity as a comprehensive evaluation indicator of link health status, and based on this, it determines link failure and adjusts aggregation strategy. Figure 1 This method is applied to communication devices equipped with WiFi7 chip modules, such as access points (APs) or client terminals (STAs). The communication devices transmit signals to target devices among multiple connected smart devices through original links. Smart devices can be mobile phones, tablets, smart home devices (fan systems, smart toilets, smart door locks, etc.).

[0049] The method includes S110 to S150:

[0050] S110, determining the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent down by the RF front end of the WiFi7 chip module.

[0051] The RF front-end (RFF) of a WiFi chip is responsible for wireless signal reception, down-conversion, and analog-to-digital conversion. Its output is a physical layer symbol stream. In OFDM systems (such as WiFi), this symbol stream consists of complex symbols across multiple subcarriers.

[0052] In S110, the numerator used to calculate the physical layer energy efficiency ratio (PEER)—effective received signal energy—is obtained. "Effective" here primarily refers to the signal component that carries useful information, eliminating the effects of noise and interference, with particular attention paid to the direct wave component that may be attenuated (e.g., through walls). Therefore, S110 preferably also includes sub-steps S101 to S104:

[0053] S101, responding to the multi-link aggregation symbol stream of the physical layer sent by the RF front end of the WiFi7 chip module.

[0054] S102 , separating a subcarrier symbol sequence corresponding to an original link from a multi-link aggregated symbol stream to obtain a symbol stream corresponding to the original link.

[0055] WiFi 7 chipsets support MLO. The physical layer symbol stream sent by the RF front-end is typically a multi-link aggregate symbol stream, which contains signals from all active links mixed together. It is necessary to separate the subcarrier symbol sequence belonging only to the original link as the target link from the multi-link aggregate symbol stream.

[0056] This relies on the Wi-Fi 7 MLO frame structure and scheduling information. The MLO protocol defines how data units are transmitted across different links (different frequency bands / channels). The chip's baseband processor or dedicated MLO processing unit knows the frequency resources (such as specific 6 GHz channels) and time slot allocations used by each link. By parsing the physical layer frame header or scheduling information, it can identify which subcarriers (or OFDM symbols) carry the data for the original link. Specifically, based on the scheduling information sent by the MAC layer (indicating the time period and frequency band / channel used for each link), the baseband processing unit performs an FFT (Fast Fourier Transform) on the ADC-sampled data to obtain frequency-domain subcarrier data. Then, based on the channel information bound to the original link, it extracts the complex symbol sequence on the subcarriers corresponding to that channel, resulting in the symbol stream corresponding to the original link.

[0057] For example, assume the original link operates on Channel 100 in the 6 GHz band. The baseband processor performs an FFT on the entire receive bandwidth (including multiple channels) and then extracts only the symbol values ​​on the subcarriers corresponding to Channel 100 to form the symbol sequence of the original link.

[0058] S103 , extracting a direct wave component from the subcarrier symbol sequence through a preset minimum mean square error channel equalizer.

[0059] The separated original link subcarrier symbol sequence is processed using a preset minimum mean square error (MMSE) channel equalizer.

[0060] During the propagation of wireless signals, multipath effects will occur (the signal reaches the receiver through different paths, generating multiple copies). The MMSE equalizer is a channel estimation and compensation technique. It estimates the frequency response of the channel (including amplitude attenuation and phase rotation) based on known pilot signals (Pilots) or training sequences (Training Sequences). Then, it applies the inverse channel response (or optimized weighting) to compensate for channel distortion, with the goal of minimizing the mean square error between the received symbol and the transmitted symbol.

[0061] When the MMSE equalizer compensates for the channel, the weight coefficients implicitly contain information about the contributions of different paths (including the direct wave and multipath reflected waves). Although it is not possible to physically separate the pure direct wave signal, for the symbol sequence after MMSE equalization, its effective components mainly reflect the signal on the strongest path (usually the direct wave). The equalization process greatly suppresses multipath interference and noise, making the symbol energy after equalization mainly represent the effective signal energy carried on the direct wave path. Therefore, the energy of the symbol sequence output by the equalizer can be regarded as an approximation of the effective received signal energy (direct wave component).

[0062] Apply the MMSE equalization algorithm to the original link subcarrier symbol sequence (the algorithm implementation is usually fixed in the baseband processing hardware or firmware of the chip) to obtain the equalized symbol sequence.

[0063] S104. Determine whether the subcarrier symbol sequence has experienced wall penetration attenuation during transmission according to the direct wave component; if the subcarrier symbol sequence has experienced wall penetration attenuation during transmission, then jump to S105 and execute S105 - S108; if the subcarrier symbol sequence has not experienced wall penetration attenuation during transmission, then jump to S108 and execute S108.

[0064] Wall penetration will cause significant additional attenuation of the signal energy. Although the MMSE equalization compensates for the channel, the average energy (or power) of the symbol sequence after equalization itself can reflect the strength of the received signal. If this energy is significantly lower than the expected received energy estimated based on the transmit power, free space path loss model, and known distance, it can be inferred that wall penetration attenuation has occurred. Specifically, calculate the average power P_rx_eq of the equalized symbol sequence within a time window and compare it with the theoretical free space received power P_rx_fs = (P_tx * G_tx * G_rx * λ²) / ((4πd)² * L) calculated based on the transmit power P_tx, wavelength λ, and distance d (if known or estimable), where G_tx and G_rx are the antenna gains and L is the system loss. If P_rx_eq << P_rx_fs (for example, more than 10 dB lower), it is determined that wall penetration attenuation has occurred.

[0065] S105 , determining the signal transmission time according to the timestamp of the direct wave component and the receiving time of the direct wave.

[0066] This is typically accomplished by measuring round-trip time (RTT) or using timestamps. The sender embeds a precise transmission timestamp, T1, in the data packet. The receiver records the reception time, T2. The receiver includes T1 and T2 in its reply packet (such as an ACK), along with its own transmission time, T3. The sender records the time it receives the reply packet, T4. The one-way transmission time can then be estimated as [(T2 - T1) + (T4 - T3)] / 2. A simpler approach is to use WiFi's Fine Timing Measurement (FTM) protocol to directly obtain distance or time information.

[0067] S106: Determine the transmission distance according to the signal transmission time and the preset transmission speed.

[0068] The transmission distance d is determined based on the transmission time and the speed of light c. For one-way time, d = (transmission time * c) / 2; if it is RTT, then d = (RTT * c) / 2.

[0069] S107: Correct the energy value of the direct wave component according to the preset wall material attenuation coefficient and transmission distance.

[0070] Knowing the distance d, the free-space path loss (PL_fs) can be estimated. The difference between the measured received power (P_rx_meas) (i.e., the symbol power after equalization) and PL_fs is the extra loss (PL_extra). If a wall penetration is detected, PL_extra is primarily attributed to wall attenuation. Based on the preset wall material attenuation coefficient (α_wall) (dB / m or dB / wall) and the wall thickness, environmental information or estimation is often required. For example, based on the distance and known layout, the number of wall penetrations (number_of_walls) and thickness can be inferred. The wall attenuation (PL_wall) can be estimated as α_wall * thickness * number_of_walls.

[0071] To more accurately reflect the direct wave energy that the receiver would receive if there were no wall attenuation (i.e., the "effective received signal energy" required for PEER calculation), the measured value P_rx_meas can be compensated: P_rx_eff = P_rx_meas + PL_wall (dB scale) or P_rx_eff = P_rx_meas * 10^(PL_wall / 10) (linear scale). This corrected P_rx_eff estimates the effective received signal energy after eliminating the attenuation of a specific wall. It better reflects the transmission potential of the link itself (assuming that the wall is absent). Specifically, based on environmental information (such as a home layout) or distance-based inference, determine the wall type (concrete, brick, wood) and approximate thickness that the signal will pass through. Query the preset α_wall table (for example, the attenuation coefficient of a concrete wall at 6 GHz is approximately 5-10 dB / 10 cm). Calculate PL_wall. Compensate P_rx_meas (dBm): P_rx_eff = P_rx_meas + PL_wall. If the environmental information is unknown, a typical value (such as 10dB) can be used conservatively for compensation.

[0072] S108: Determine the effective received signal energy according to the direct wave component.

[0073] If wall attenuation does not occur, the average power P_rx_meas of the equalized symbol sequence is directly used as the effective received signal energy P_rx_eff; if wall attenuation occurs, the average power P_rx_meas of the compensated symbol sequence is used as the effective received signal energy P_rx_eff.

[0074] Traditional BER detection methods focus solely on correct decoding, ignoring the absolute value and composition of the received signal energy. This step uses MMSE equalization to extract the direct wave component energy, combined with wall penetration detection and compensation, to more accurately quantify the received strength of the effective physical layer signal, providing a reliable basis for evaluating the physical layer transmission efficiency (PEER). This helps distinguish whether transmission failures are caused by inherent signal weakness (low PEER) or other factors (such as congestion). It also obtains the effective received signal energy, a key parameter characterizing physical layer signal transmission efficiency, while minimizing the additional attenuation caused by environmental factors (such as wall penetration). This allows subsequent PEER calculations to better reflect the physical characteristics of the link itself, achieving a combination of advanced channel equalization techniques and environmental awareness (wall penetration detection and compensation), surpassing assessments based solely on BER or simple RSSI (Received Signal Strength Indicator).

[0075] S120, counting the total number of application layer valid data bytes confirmed by all TCP confirmation messages corresponding to the original link in the transmission layer of the WiFi7 chip module within a preset time window to obtain the valid data volume.

[0076] Obtain the numerator used to calculate the transport layer throughput efficiency ratio (TTER) - the effective data volume. This is the total amount of application layer data successfully received and acknowledged by the target device through the original link.

[0077] The preset time window is a defined statistical period, such as 100ms, 500ms, or 1s. The window size requires a balance between real-time performance and statistical stability and can be set based on actual usage needs.

[0078] In the TCP protocol, the receiving end (destination device) sends a TCP acknowledgment packet (ACK packet) to the sending end (this WiFi 7 device) every time it successfully receives a certain amount of data (or according to an algorithm). This ACK number indicates the total number of valid data bytes acknowledged by the receiving end (the number of bytes between the initial sequence number and the ACK number).

[0079] In an MLO scenario, the data stream of a TCP connection may be split and transmitted across multiple links. The TCP ACK returned by the receiving end is for the entire TCP connection and does not specify which link the data is being acknowledged on. It is necessary to determine whether a particular ACK message is generated in response to data sent over the original link. Therefore, preferably, the TCP acknowledgment message includes five-tuple data; the TCP acknowledgment message corresponding to the original link in S120 is determined through S118-S119:

[0080] S118 , determining the link to which the TCP confirmation message belongs through the five-tuple data of the TCP confirmation message of the transport layer.

[0081] Quintuple, that is, a TCP connection is uniquely identified by a quintuple: <source IP address, source port, destination IP address, destination port, protocol (TCP)>.

[0082] This Wi-Fi 7 device (as the sender) knows which physical link it sent which packets. When it receives a TCP ACK packet, it first parses the IP and TCP headers of the ACK packet to extract the five-tuple information. Then, based on this five-tuple, it searches the local connection tracking table (CTT) or MLO scheduling mapping table. This table records the physical links over which the packets of this TCP connection were sent. More specifically, it records which packets (sequence number ranges) were sent over the original link.

[0083] S119: Determine the TCP acknowledgment message whose link is the original link as the TCP acknowledgment message corresponding to the original link.

[0084] The TCP ACK message whose link is the original link is identified as the TCP ACK message corresponding to the original link. The TCP ACK number sent by the receiver confirms the highest sequence number of consecutive data received in order. Therefore, the data acknowledged by an ACK message may originate from multiple links (if data is received out of order, the ACK number is only confirmed up to the highest point in the consecutive sequence). To accurately count the amount of data sent and successfully acknowledged via the original link, the following steps are required: when a TCP ACK message is received, based on its ACK number A, query the local record to find all data packets with sequence numbers less than or equal to A that were sent via the original link; and accumulate the number of data bytes in these packets that have not been previously acknowledged by ACKs. This accumulated value represents the amount of new valid data acknowledged by this ACK message for the original link.

[0085] Specifically, a state is maintained for each TCP connection and each link: Last_ACKed_Seq[link] records the highest sequence number last acknowledged on the link. When a packet with ACK number A is received and determined to be associated with the original link (using the quintuple and link mapping): all packets sent via the original link with sequence numbers in the range (Last_ACKed_Seq[orig], A]) are found; the payload bytes in these packets (excluding the TCP / IP header) are accumulated to obtain the additional valid data ΔBytes_eff acknowledged by the ACK packet for the original link; Last_ACKed_Seq[orig] is updated to A; and ΔBytes_eff is added to the total valid data in the current time window, Total_Bytes_eff.

[0086] The effective data volume is calculated by accumulating the newly added effective data volume ΔBytes_eff confirmed by all TCP ACK messages identified as "corresponding to the original link" within a preset time window to obtain the effective data volume Bytes_eff of the window.

[0087] This step obtains the key parameter that characterizes the effective throughput efficiency of the transport layer - the effective data volume. It provides the core input for calculating TTER and the final entropy capacity, and directly solves the core defect of the BER detection method that cannot perceive the effectiveness of the application layer. By accurately counting the amount of data actually successfully received by the application layer (confirmed by TCP), it directly reflects the true effectiveness of the link at the transport layer and application layer, avoiding the mistake of mistaking the link for availability when the TCP window is closed (effective throughput is 0).

[0088] By utilizing TCP ACK messages and five-tuple information, combined with MLO scheduling mapping, the transport layer confirmation information is accurately associated with the specific physical link, achieving cross-layer (transport layer to physical layer) information fusion.

[0089] S130: Request the target device for the total amount of data sent and the physical layer transmission energy within the same time window.

[0090] Obtain the denominator for calculating PEER (physical layer transmission energy) and TTER (total amount of data sent).

[0091] At the end of each preset time window, the Wi-Fi 7 device sends a request message to the target device. This message requests the target device to provide: the total amount of data sent (Bytes_sent), which is the total number of application layer data bytes sent by the target device over the original link during the time window (typically, the total payload of the data packets sent by the target device when acting as a TCP sender); and the physical layer transmit energy (E_tx or P_tx_avg), which is the average transmit power or total transmit energy of the signal sent by the target device over the original link during the time window. This typically refers to the output power of the RF power amplifier.

[0092] This can be sent to the target device via a dedicated management frame or by leveraging existing protocol extensions (e.g., by carrying the request information in a data frame). The request message should include the start and end timestamps of the time window (or window ID) and the requested link identifier (the original link).

[0093] After receiving the request, the target device queries its local records and counts the total number of application layer data bytes (Bytes_sent) sent through the specified original link within the specified time window. It also calculates the average transmit power (P_tx_avg) or total transmit energy (E_tx = P_tx_avg * window duration) of the signal sent through the specified original link within the specified time window. Bytes_sent and P_tx_avg (or E_tx) are encapsulated in a response message and sent back to the local WiFi 7 device.

[0094] After receiving the response, the device extracts Bytes_sent and P_tx_avg (or E_tx).

[0095] S140 , determining the entropy capacity corresponding to each time window according to the ratio of the effective received signal energy to the physical layer transmission energy, and the ratio of the effective data volume to the total amount of transmitted data.

[0096] Physical layer energy efficiency ratio (PEER): PEER = effective received signal energy / physical layer transmitted energy. PEER measures how much of the transmit energy expended by the transmitter is effectively captured by the receiver (primarily referring to the signal component carrying valid information, such as the direct wave). It reflects the efficiency of physical layer signal transmission. A low PEER indicates significant signal loss during spatial transmission (such as severe wall attenuation and path loss) or signal overwhelmed by interference. Even if the receiver receives the signal, the signal-to-noise ratio (SNR) may be low, resulting in low high-layer transmission efficiency.

[0097] Transport layer throughput efficiency ratio (TTER) is calculated as: TTER = effective data volume / total data volume sent. TTER measures how much of the total data volume sent by the sender is successfully received and acknowledged by the receiver (typically, the number of bytes confirmed by TCP ACKs). It directly reflects the actual effective throughput efficiency of the transport layer (e.g., TCP). A low TTER (especially one close to 0) indicates that a large number of data packets are lost or cannot be acknowledged by upper layers, preventing the application layer from receiving valid data.

[0098] Entropy capacity (EC), where EC = f(PEER, TTER), combines information from the physical and transport layers to more accurately reflect the end-to-end communication capabilities of the original link. If PEER is high but TTER is low (a typical characteristic of a pseudo-active link), EC will be very low, indicating that the physical layer signal is acceptable but the transport layer is completely inaccessible. If both PEER and TTER are low, EC will also be low, indicating that poor physical layer signaling has led to transmission failures. EC is high only when both PEER and TTER are high, indicating a truly healthy link. BER only focuses on bit errors at the PHY layer and cannot perceive the effectiveness of the transport and application layers. Entropy capacity (EC), combined with TTER, directly captures the success or failure of valid data transmission at the application layer, enabling accurate identification of pseudo-active links that BER detection methods cannot detect.

[0099] By combining the efficiency information of the physical and transport layers, we calculate the core metric of this solution—entropy capacity (EC)—for a comprehensive assessment of link health. The physical layer energy efficiency ratio (PEER) is calculated as follows: PEER = P_rx_eff / P_tx_avg. (If power is used, the ratio is dimensionless, typically between 0 and 1, but can be greater than 1, for example, when the receiver has a high-gain antenna. It is more commonly expressed in dB: PEER_dB = P_rx_eff_dBm - P_tx_avg_dBm.) Alternatively, PEER = E_rx_eff / E_tx (if energy is used).

[0100] A PEER close to 1 (or 0 dB) indicates that almost all transmitted energy is effectively received, indicating high efficiency. A very low PEER indicates that significant signal loss occurs during transmission.

[0101] Calculate the transport layer throughput efficiency ratio (TTER): TTER = Bytes_eff / Bytes_sent. A TTER close to 1 indicates that almost all sent data was successfully received and acknowledged, indicating high throughput efficiency. A TTER close to 0 (especially 0) indicates that almost all sent data was lost or unacknowledged, indicating a transport layer failure.

[0102] Calculate entropy capacity (EC): EC = f(PEER, TTER). The specific function form can be designed in various ways: the product form, EC = PEER * TTER, is the most intuitive form. It requires both high PEER and high TTER for high EC, combining physical signal transmission efficiency and effective data transmission efficiency. The weighted sum form, EC = α * PEER + β * TTER (α + β = 1), can adjust the weights based on the scenario (for example, increasing β when throughput is more important). The logarithmic form (e.g., capacity), EC = log2(1 + SNR_eff) * TTER, where SNR_eff is the equivalent signal-to-noise ratio estimated from P_rx_eff and noise power. This is closer to the concept of information-theoretic capacity but is more computationally complex.

[0103] At the end of each time window, an EC value is calculated, which represents the comprehensive communication capability of the original link within the window.

[0104] EC addresses the core flaw of BER detection methods—its inability to identify "pseudo-active links." When the physical layer is active (PEER may be acceptable) but the transport layer is inoperative (TTER = 0), EC = 0 (or very low), enabling accurate identification. Similarly, transmission failures (low TTER) caused by poor physical layer performance (low PEER) can also result in low EC. EC goes beyond single-layer (PHY or Transport) metrics, enabling cross-layer awareness and providing a more comprehensive and accurate assessment of link status. This is particularly advantageous in identifying "pseudo-active links."

[0105] S150: If the entropy capacity corresponding to a preset number of consecutive time windows is less than a preset first entropy capacity threshold, the original link is determined as a failed link.

[0106] Furthermore, S151-S152 can be used to eliminate the interference of congestion on the determination of whether the link is failed. Preferably, S150 further includes sub-steps S151-S152:

[0107] S151: Determine whether a network congestion flag appears at the transport layer according to the ECN field in the header of the IP message associated with the transport layer.

[0108] ECN (Explicit Congestion Notification) is a mechanism of the IP protocol used to notify endpoints when network congestion occurs. When a router supporting ECN detects congestion, it sets the ECN field in the IP header (usually set to CE, Congestion Experienced). After the receiving end (this device) receives an IP packet (belonging to the target TCP connection) containing the ECN-CE mark, it sets the ECN-Echo flag in the returned TCP ACK packet to inform the sending end (the target device) that congestion has occurred. The sending end (the target device) then responds by setting the CWR (Congestion Window Reduced) flag in the next data packet.

[0109] When this device (as the receiving end) receives a data packet (IP packet) sent by the target device, it checks the ECN field in its IP header. If the ECN-CE mark is found, it records that the network congestion flag has occurred for this connection / link. Or when receiving a TCP ACK packet (responding to the data sent by this device) from the target device, it checks whether the ECN-Echo flag in its TCP header is set (which means the target device, as the receiving end, has received an ECN-CE packet).

[0110] S152, if there is no network congestion in the transport layer and the entropy capacity corresponding to a continuous preset number of time windows is less than a preset first entropy capacity threshold, then the original link is determined to be a failed link.

[0111] Before determining that the link has failed, first check whether the low EC is caused by network congestion (rather than a local wireless link problem). If no ECN congestion mark has been detected for the TCP connection associated with the original link within a recent period of time (such as the time covering these consecutive N windows), that is, no ECN-CE packet or ECN-Echo ACK has been received, then it can be excluded that network congestion is the main cause of the low TTER (and thus the low EC). At this time, the low EC is more likely to be "pseudo-active" or truly deteriorated due to physical problems (such as severe attenuation, interference) of the original link itself, thus confirming the link failure.

[0112] If a congestion mark is detected, even if EC < EC_threshold1 and this occurs continuously for N times, do not immediately determine that the link has failed. Because the low performance may be caused by congestion on the network path, and switching the wireless link may not solve the problem. At this time, other congestion control mechanisms can be triggered or only the count_low_ec can be increased without triggering the failure determination (or use a looser threshold).

[0113] Through S151~S152, the misjudgment caused by network core congestion (misjudging non-link problems as link failures) can be significantly reduced, and the accuracy of failure judgment can be improved.

[0114] For stopping aggregation of failed links in MLO, preferably, after S150, the method further includes S162-S163:

[0115] S161: Generate a link failure signal, which includes the link identifier of the original link and a MAC close operation code.

[0116] Prepare instructions to notify the target device to close the failed link. The link failure signal is a customized management frame or uses the extended field of an existing management frame (such as the Action Frame) to standardize the message format of the link failure notification. It includes the following contents: Link Identifier, which uniquely identifies the original link to be closed (for example, the frequency band used, channel number, MLO link ID); MAC Disable Opcode, a predefined code that instructs the receiving end to perform the "disable MAC layer resource allocation for the specified link" operation.

[0117] S162, shielding the resource allocation request of the original link at the local MAC layer of the WiFi7 chip module, and selecting a target link from multiple communication links of the WiFi7 chip module, where the target link is different from the original link.

[0118] The failed link is immediately stopped locally, and a backup link is selected for notification to the target device and subsequent communication. The WiFi chip's MAC layer is responsible for link scheduling and resource allocation (such as time slots and channel access). The MAC layer's scheduling policy or link status table can be modified to implement local MAC layer shielding. For example, the original link status can be marked as "failed" or "disabled." Afterwards, when performing multi-link aggregation scheduling, the MAC layer ignores the original link's resource requests and no longer allocates any data streams (including management frames and data frames) to this link for transmission. The link's RF front end may enter a low-power state or monitor mode.

[0119] Immediately stop sending data to the failed link, avoiding protocol overhead waste and freeing up resources for other healthy links.

[0120] Selecting a target link requires choosing an available link different from the original link (for example, if the original link is 6 GHz, choose 5 GHz or 2.4 GHz as the target link) to send the link failure signal. A simple strategy is to select another link with the highest current EC or the best status. Alternatively, the selection can be based on link load, historical performance, and other factors. For example, the MAC layer or MLO management module excludes the original link from the active link list and then selects a target link based on a preset strategy (such as the link with the highest EC) as the Target_Link to ensure that the failure notification is reliably delivered to the target device.

[0121] S163: Send a link failure signal to the target device via the target link, so that the target device closes the MAC layer resource allocation of the original link according to the link identifier in response to the MAC close operation code.

[0122] The target device is notified to collaboratively disable the MAC function of the failed link, the link failure signal generated in S161 is encapsulated into a frame, and sent to the target device through the target link Target_Link selected in S162.

[0123] After receiving the link failure signal, the target device parses the link identifier and MAC close operation code, and the MAC layer of the target device performs the following operations: stops sending any data (management frames, data frames) on the original link; stops listening to resource allocation (such as TXOP) on the original link; may put the RF front end of the original link into a low power state; updates its MLO link status table and marks the link as "remote disabled".

[0124] It achieves bilateral closure of failed links, maximizes resource savings, improves the efficiency of aggregation of remaining healthy links, ensures that both communicating parties stop using the failed link simultaneously, and completely avoids resource waste and interference caused by either party continuing to try to communicate on the link.

[0125] After the original link is blocked, it is periodically detected whether it is restored to be available. Preferably, after S163, the method further includes S171 to S173:

[0126] S171: Send a test data stream to a target device via an original link at a preset time interval.

[0127] The time interval is preset, for example, every 5 seconds, 10 seconds, 30 seconds or 5 minutes; the test data stream is data specifically used to detect the link status.

[0128] Preferably, the test data stream includes a UDP probe packet, and the UDP probe packet includes a forced response identifier, so that the target device returns a TCP confirmation message after detecting the forced response identifier.

[0129] UDP probe packets use the connectionless, unreliable UDP protocol to send small data packets. This has low overhead and does not rely on the complex retransmission and confirmation mechanisms of TCP. It can quickly test the basic connectivity and transmission capacity of the link.

[0130] The UDP probe packet's payload contains a special Mandatory Response Identifier (MRI). This identifier indicates that the target device must return a response packet, typically a TCP ACK packet, after receiving this packet. See S172 for explanation. The MRI can be a specific magic number or a protocol-defined flag.

[0131] Specifically, at the detection moment, the local MAC layer temporarily lifts the sending shield on the original link (the receiving may still be blocked), constructs one or more UDP detection packets, sets the target IP / port (a listening port or agreed port on the target device), includes MRI in the payload, and sends these UDP detection packets through the original link. After the sending is completed, the sending shield on the original link is restored (waiting for a response), so as to actively detect the recovery of the failed link with low overhead.

[0132] S172: Determine the entropy capacity corresponding to the test data stream.

[0133] The probe process essentially conducts a small-scale, targeted communication test. The process from S110 to S140 can be applied to this test to calculate the probe entropy capacity EC_probe. Specifically, the effective received signal energy (P_rx_eff_probe) is the received energy obtained by executing S110 while receiving the response packet from the target device. The key to determining the effective data volume (Bytes_eff_probe) lies in triggering a response from the target device. Since the target device receives a UDP probe packet (containing an MRI), it constructs and returns a TCP ACK packet based on the MRI (even if there is no corresponding TCP data stream). This ACK packet can acknowledge a virtual sequence number (e.g., 0). Alternatively, after receiving the MRI, the target device proactively sends a small TCP data packet (e.g., 1 byte) to the local device. Upon receiving this TCP data packet, the local device returns a TCP ACK packet to the target device. The local device counts the number of bytes acknowledged by the returned ACK packet (i.e., the number of bytes in the small TCP packet sent by the target device) as Bytes_eff_probe.

[0134] The total amount of data sent (Bytes_sent_probe) is the total number of bytes in all UDP probe packets sent during this probe (application layer payload). The physical layer transmit energy (P_tx_avg_probe or E_tx_probe) records the average transmit power or total energy of the device when sending probe packets (downlink). Request or record the transmit energy of the target device when sending response packets (uplink, optional if calculating uplink PEER).

[0135] Calculate PEER_probe and TTER_probe: Downlink PEER: P_rx_eff_probe (target device sends, local device receives) / P_tx_avg_target_probe (target device sends) (provided by the target device) or Uplink PEER: P_rx_eff_probe (local device receives responses) / P_tx_avg_local_probe (local device sends probes). Since this is a probe, you can focus on TTER_probe (amount of successfully received response data / amount of sent probe data) or directly determine whether a valid response was received. Alternatively, simply use the received response signal energy as a simplified physical layer metric.

[0136] The entropy capacity EC_probe can be calculated using a simplified calculation, such as EC_probe = TTER_probe (if a successful response is received, TTER_probe = 1) or combined with a simplified PEER estimate. The core is to determine whether the link can complete a valid two-way communication.

[0137] S173: If the entropy capacity corresponding to the test data flow is greater than a preset second entropy capacity threshold, restoring the resource allocation authority of the original link, wherein the second entropy capacity threshold is less than the first entropy capacity threshold.

[0138] The second entropy capacity threshold, EC_threshold2, is set lower than the first threshold, EC_threshold1. For example, EC_threshold1 is set to 0.05 and EC_threshold2 is set to 0.03. Because the amount of probe data is small, statistics are less stable than those in a regular window, and initial recovery performance may be suboptimal. Setting a lower threshold allows for more proactive link reactivation.

[0139] If the calculated probe entropy capacity EC_probe > EC_threshold2, at the local MAC layer of this device, lift the resource allocation shielding for the original link, and allow MLO to reschedule data to this link again; if EC_probe <= EC_threshold2, then maintain the shielding state of the original link and wait for the next probe, implementing an automatic recovery mechanism for the failed link, enabling the MLO system to dynamically adapt to changes in the wireless environment and maximize the utilization of available link resources.

[0140] Preferably, before S150, the method further includes S149: In S149, if the entropy capacity corresponding to a certain time window is less than a preset third entropy capacity threshold, determine the original link as a failed link, where the third entropy capacity threshold is less than the second entropy capacity threshold.

[0141] In S149, a fast failure determination mechanism is provided to handle the situation where the link is suddenly completely interrupted (such as when the device is moved out of the coverage area or the link is briefly interfered when the microwave oven is started).

[0142] The third entropy capacity threshold (EC_threshold3) sets a very low threshold (such as 0.01, 0.001), even close to 0. If the calculated EC < EC_threshold3 for a certain time window, immediately determine that the original link has failed, without waiting for N consecutive windows. Directly jump to S161 to execute the failure handling process.

[0143] Applicable scenarios: When the link is suddenly completely interrupted (such as the target device is powered off, moved out of the coverage area, or severely blocked and interfered), resulting in TTER ≈ 0 and PEER being extremely low (no signal or strong interference), the EC will instantaneously drop to an extremely low value. At this time, waiting for N consecutive windows to determine failure is too slow. S149 provides the ability for fast response.

[0144] S149 is a supplement to S150. S150 handles the situation of continuous poor performance, and S149 handles the situation of instantaneous complete interruption, with EC_threshold3 < EC_threshold2 < EC_threshold1, to improve the response speed to sudden and catastrophic link failures.

[0145] The overall processing flow includes S1~S6:

[0146] S1, initialization processing, set the time window size, thresholds (EC_threshold1, EC_threshold2, EC_threshold3, N), and initialize the counter.

[0147] S2. Loop through each time window. In S110, obtain / calculate P_rx_eff of the current window's original link (including sub - steps S101 - S108: separating symbol streams, MMSE equalization, wall - penetration judgment and correction, etc.). In S120, count Bytes_eff of the current window's original link (including sub - steps S118 - S119: associating TCP ACK through quintuple and link mapping). In S130, request and obtain Bytes_sent and P_tx_avg (or E_tx) of the target device in the same window. In S140, calculate PEER and TTER, and then calculate EC of the current window. In S149, perform a quick failure check. If EC < EC_threshold3, immediately determine failure and jump to S161. In S150, perform a continuous failure check. Optionally execute S151 to check the ECN congestion flag of the associated TCP connection. Optionally execute S152. If there is no congestion and EC < EC_threshold1, jump to S161. If there is an ECN congestion flag, wait for the ECN congestion flag to disappear and then loop through the restored normal time window.

[0148] S3. Post - failure processing. In S161, generate a link failure signal containing the link ID and MAC shutdown opcode. In S162, locally mask the original link resource allocation, select the target link. In S163, send the link failure signal to the target device through the target link to make the target device close the original link MAC resources.

[0149] S5. Link recovery detection processing. In S171, send a UDP probe packet containing the mandatory response identifier MRI through the original link at a preset interval. In S172, determine EC_probe of this detection. In S173, if EC_probe > EC_threshold2, restore the local original link resource allocation.

[0150] S6. Return to S2.

[0151] In the embodiment of this application, the throughput efficiency ratio of the transport layer is characterized by the ratio of the effective received signal energy to the physical - layer transmitted energy, and the energy efficiency ratio of the physical layer is characterized by the ratio of the effective data volume to the total transmitted data volume. The obtained entropy capacity can synthesize the data of the physical layer and the transport layer, truly reflect the communication ability of the original link, and make a relatively accurate determination of whether the original link fails. Thus, the WIFI7 module can abandon the aggregation of the failed original link and improve the aggregation performance of MLO.

[0152] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0153] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-device synchronous communication method for a WiFi7 chip module, characterized in that: The method is applied to a communication device equipped with a WiFi7 chip module, wherein the communication device transmits signals to a target device among a plurality of connected smart devices via an original link, and the method includes: Determine the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the radio frequency front end of the WiFi7 chip module; Counting the total number of valid data bytes of the application layer confirmed by all TCP confirmation messages corresponding to the original link at the transport layer of the WiFi7 chip module within a preset time window to obtain the valid data volume; Requesting the target device for a total amount of data to be sent and a physical layer transmission energy within the same time window; Determining the entropy capacity corresponding to each time window according to the ratio of the effective received signal energy to the physical layer transmit energy, and the ratio of the effective data volume to the total amount of transmitted data; If the entropy capacity corresponding to a preset number of consecutive time windows is less than a preset first entropy capacity threshold, the original link is determined to be a failed link.

2. The method according to claim 1, wherein Before determining the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the radio frequency front end of the WiFi7 chip module, the method further includes: In response to the multi-link aggregation symbol stream of the physical layer issued by the radio frequency front end of the WiFi7 chip module; A subcarrier symbol sequence corresponding to the original link is separated from the multi-link aggregated symbol stream to obtain a symbol stream corresponding to the original link.

3. The method according to claim 2, wherein The determining of the effective received signal energy according to the symbol stream corresponding to the original link in the physical layer sent by the radio frequency front end of the WiFi7 chip module includes: Extracting a direct wave component from the subcarrier symbol sequence by a preset minimum mean square error channel equalizer; The effective received signal energy is determined according to the direct wave component.

4. The method according to claim 3, wherein Before determining the effective received signal energy according to the direct wave component, the method further includes: determining, according to the direct wave component, whether the subcarrier symbol sequence experiences wall attenuation during transmission; If yes, determining the signal transmission time according to the timestamp of the direct wave component and the reception time of the direct wave; Determining a transmission distance based on the signal transmission time and a preset transmission speed; The energy value of the direct wave component is corrected according to the preset wall material attenuation coefficient and the transmission distance.

5. The method according to claim 4, wherein The TCP confirmation message includes five-tuple data; before counting the total number of application layer valid data bytes confirmed by all TCP confirmation messages corresponding to the original link of the transport layer of the WiFi7 chip module within a preset time window to obtain the valid data amount, the method further includes: Determine the link to which the TCP confirmation message belongs through the five-tuple data of the TCP confirmation message of the transport layer; The TCP acknowledgment message whose link is the original link is determined as the TCP acknowledgment message corresponding to the original link.

6. The method according to claim 5, wherein After determining the original link as a failed link if the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, the method further includes: Generate a link failure signal, wherein the link failure signal includes a link identifier and a MAC close operation code of the original link; Shielding the resource allocation request of the original link at the local MAC layer of the WiFi7 chip module, and selecting a target link from multiple communication links of the WiFi7 chip module, where the target link is different from the original link; The link failure signal is sent to the target device via the target link, so that the target device closes the MAC layer resource allocation of the original link according to the link identifier in response to the MAC close operation code.

7. The method according to claim 6, wherein After sending the link failure signal to the target device through the target link so that the target device closes the MAC layer resource allocation of the original link according to the link identifier in response to the MAC close operation code, the method further includes: Sending a test data stream to the target device via the original link at a preset time interval; determining an entropy capacity corresponding to a test data stream; If the entropy capacity corresponding to the test data flow is greater than a preset second entropy capacity threshold, the resource allocation authority of the original link is restored, wherein the second entropy capacity threshold is less than the first entropy capacity threshold.

8. The method according to claim 7, wherein The test data stream includes a UDP probe packet, and the UDP probe packet includes a forced response identifier, so that the target device returns a TCP confirmation message after detecting the forced response identifier.

9. The method according to claim 8, wherein Before determining the original link as a failed link if the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, the method further includes: If the entropy capacity corresponding to a certain time window is less than a preset third entropy capacity threshold, the original link is determined as a failed link, wherein the third entropy capacity threshold is less than the second entropy capacity threshold.

10. The method according to claim 9, wherein If the entropy capacity corresponding to a consecutive preset number of time windows is less than a preset first entropy capacity threshold, determining the original link as a failed link further includes: Determine whether a network congestion flag appears at the transport layer according to an ECN field in a header of an IP message associated with the transport layer; If there is no network congestion at the transport layer and the entropy capacity corresponding to a preset number of consecutive time windows is less than a preset first entropy capacity threshold, the original link is determined to be a failed link.

Images