A method and system for multi-mode converged wireless communication chip

Abstract

This invention discloses a method and system for a multi-mode fusion wireless communication chip. By constructing a four-dimensional joint observation vector set, the received signals of different communication systems are regarded as observations of the same channel under different projection spaces. Then, the local coherent structure is extracted and piecewise fitting is performed on the four-dimensional complex unit sphere to generate a piecewise constant joint channel state identifier sequence. This sequence does not depend on pilot symbols and is derived only from the consistency of the received signal itself, possessing low overhead, fast response, and strong cooperative characteristics. On this basis, resource allocation, parameter configuration, phase preset, interference cancellation, and soft information fusion are all carried out around the same joint channel state identifier, making the originally fragmented reception processes of each system form a closed-loop linkage. Therefore, it achieves a significant reduction in channel state representation overhead, close synchronization between state updates and actual channel changes, and deep cooperation between the physical layer and demodulation layer of multiple systems.

Description

Technical Field

[0001] This invention relates to the field of wireless transmission technology, specifically to a method and system for a multi-mode fusion wireless communication chip. Background Technology

[0002] Currently, multimode wireless communication chips generally employ an independent operation approach for each communication system: narrowband IoT, wideband orthogonal frequency division multiplexing, frequency hopping spread spectrum, and pulse ultra-wideband receiver links each have their own clock sources, acquire signals, estimate channels, allocate resources, and make demodulation decisions. To support channel state acquisition for different systems, dedicated pilot sequences are typically inserted into each system, and independent channel state information is generated by separate channel estimation modules. Each system then performs resource scheduling and parameter configuration independently based on its own estimation results. This design results in pilot overhead increasing linearly with the number of systems, which is particularly prominent in environments with rapidly changing channel states. When mobile terminals travel through urban canyons or operate at high speeds, the multipath structure changes significantly within milliseconds. Due to long pilot intervals, large estimation delays, and separate feedback paths, each system struggles to synchronously perceive the consistent nature of channel evolution.

[0003] This raises a clear technical problem: different communication systems present fragmented modeling of their responses to the same dynamic channel, making it impossible to achieve joint representation by leveraging their inherent correlations. This results in high overhead for channel state acquisition, delayed updates, and a lack of cross-system coordination capabilities, ultimately restricting transmission reliability and resource utilization efficiency in scenarios where multiple systems coexist. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for a multi-mode fusion wireless communication chip, which achieves a significant reduction in channel state representation overhead, close synchronization between state updates and actual channel changes, and deep collaboration between the physical layer and demodulation layer of multiple systems, thereby effectively improving the transmission reliability and resource scheduling efficiency of multi-system coexistence systems in environments with rapidly changing channels.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a method for a multi-mode fusion wireless communication chip, comprising:

[0006] Under the control of a unified time base unit, resampling and carrier frequency offset compensation are performed on the received signals of different communication systems to construct a four-dimensional joint observation vector set;

[0007] Extract the local coherence structure of the four-dimensional joint observation vector along the time axis, calculate the principal coherence direction within each time window, and form a time-varying joint feature trajectory;

[0008] Segmented fitting is performed on the joint feature trajectory on a four-dimensional complex unit sphere to generate a segmented constant joint channel state identifier sequence. Based on the joint channel state identifier sequence, the resource allocation weights of different communication systems in each scheduling period are calculated to generate a time-frequency resource allocation mapping.

[0009] Based on the time-frequency resource allocation mapping, the physical layer parameter configuration combination of each communication system within the scheduling cycle is determined to form a joint operating state tuple. Based on the joint operating state tuple, the local oscillator phase trajectory and sampling clock phase offset of each receiving link are synchronously adjusted.

[0010] Using the sampled values ​​of each receiving link alignment time within the same scheduling period, a cross-system interference cancellation reference signal is constructed and injected into the corresponding receiving path. The interference-cancelled received signals are sent to a shared joint decoding unit. Based on the joint channel state identifier, the demodulation soft information of each path is weighted and fused, and symbol-level decision is performed.

[0011] Preferably, the construction of the four-dimensional joint observation vector set includes:

[0012] The reference clock output from the master oscillator drives the receiving links of each communication system.

[0013] The received signals of each communication system are resampled to a common time grid, and the corresponding carrier frequency offset compensation is applied to the resampled signals.

[0014] The compensated four signals are combined into a four-dimensional joint observation vector.

[0015] Preferably, the extraction of the local coherence structure of the four-dimensional joint observation vector along the time axis includes:

[0016] Calculate the sliding energy ratio of the received signals of each communication system, identify the energy jump point, and generate an adaptive time window centered on the energy jump point;

[0017] Calculate the principal coherence direction for the joint observation vector matrix within each time window, and stack all principal coherence directions in chronological order to form a joint feature trajectory.

[0018] Preferably, the step of performing piecewise fitting of the joint feature trajectory on a four-dimensional complex unit sphere includes:

[0019] Calculate the spherical angle between three adjacent points of the joint feature trajectory, identify the inflection point, divide the trajectory into segments with the inflection point as the boundary, and calculate the centroid direction of each segment;

[0020] Verify whether the average spherical distance from each segment point to the centroid direction meets the preset threshold;

[0021] Each sampling point is mapped to the corresponding trajectory segment index to generate a joint channel state identifier sequence.

[0022] Preferably, the generation of the time-frequency resource allocation map includes:

[0023] Calculate the four-dimensional modulus distribution vector corresponding to each joint channel state identifier, and define the resource allocation weight vector based on the modulus distribution vector;

[0024] The joint channel state identifier that appears most frequently in each scheduling cycle is statistically analyzed, and the resource allocation weights are mapped to the time-frequency resource block allocation ratios, and the total amount is normalized.

[0025] Preferably, determining the physical layer parameter configuration combination of each communication system within the scheduling period based on the time-frequency resource allocation mapping includes:

[0026] A finite number of physical layer operating states are predefined for each communication system, and the information carrying capacity per unit resource block and the average duration of each operating state are calibrated.

[0027] Based on the number of allocated resource blocks and the scheduling cycle length, select the running state that meets the conditions;

[0028] Combine the four selected running states into a joint running state tuple.

[0029] Preferably, the synchronous adjustment of the local oscillator phase trajectory and sampling clock phase offset of each receiving link includes:

[0030] Store the corresponding local oscillator phase initial value vector for each joint running state tuple;

[0031] Store the corresponding sampling clock phase offset vector for each joint running state tuple;

[0032] Before the start of the scheduling period, the corresponding phase value is written to each receiving link, and the ideal sampling time set of each receiving link within the scheduling period is defined.

[0033] Preferably, the construction of the cross-system interference cancellation reference signal includes:

[0034] Extract the sampled values ​​of the four received signals at the alignment time within the same scheduling period;

[0035] Apply a sliding window processing to the sampled value sequence and calculate the projection matrix within the window;

[0036] The interference transfer matrix is ​​calculated based on the projection matrix, and a cancellation signal is generated and injected into the receiving path based on the interference transfer matrix.

[0037] Preferably, the execution of the symbol-level decision includes:

[0038] Each communication system outputs symbol-level soft information, and obtains a confidence vector based on the joint channel state identifier;

[0039] Each soft information is weighted according to its confidence vector, and the weighted soft information is aligned according to its bit position before a symbol-level fusion decision is performed.

[0040] On the other hand, this invention proposes a system for a multi-mode fusion wireless communication chip, comprising a unified time base unit, a signal processing unit, and a joint decoding unit, wherein:

[0041] A unified time base unit controls the synchronous sampling of signals from different communication systems across all receiving links;

[0042] The signal processing unit is used to construct a four-dimensional joint observation vector set, extract local coherent structures to form joint feature trajectories, and generate a joint channel state identifier sequence by segmental fitting on a four-dimensional complex unit sphere.

[0043] The signal processing unit is also used to calculate resource allocation weights based on the joint channel state identifier sequence, generate time-frequency resource allocation mappings, and determine the physical layer parameter configuration combinations for each communication system.

[0044] The signal processing unit is also used to adjust the local oscillator phase and sampling clock phase of each receiving link according to the physical layer parameter configuration, and to construct and inject cross-system interference cancellation reference signal;

[0045] The joint decoding unit receives the received signals after interference cancellation, weights and fuses the demodulated soft information based on the joint channel state identifier, and performs symbol-level decision-making.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] This invention constructs a four-dimensional joint observation vector set, treating received signals from different communication systems as observations of the same channel in different projection spaces. It then extracts the local coherent structure and performs piecewise fitting on a four-dimensional complex unit sphere to generate a piecewise constant joint channel state identifier sequence. This sequence is independent of pilot symbols and is derived solely from the consistency of the received signal itself, exhibiting low overhead, fast response, and strong coordination characteristics. Based on this, resource allocation, parameter configuration, phase presetting, interference cancellation, and soft information fusion all revolve around the same joint channel state identifier, forming a closed-loop linkage between the previously fragmented reception processes of different systems. Therefore, it achieves a significant reduction in channel state representation overhead, close synchronization between state updates and actual channel changes, and deep coordination between the physical layer and demodulation layer of multiple systems, thereby effectively improving the transmission reliability and resource scheduling efficiency of multi-system coexistence systems in environments with rapidly changing channels. Attached Figure Description

[0048] Figure 1 This is a flowchart of the method for using the multi-mode fusion wireless communication chip of the present invention;

[0049] Figure 2This is a block diagram of the multi-mode fusion wireless communication chip system of the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] like Figure 1 As shown, this invention proposes a method for a multi-mode fusion wireless communication chip. In environments with rapidly changing channel states, it utilizes the inherent correlation between received signals from different communication systems to construct a low-overhead joint channel state characterization method to support reliable transmission scheduling through multi-system collaboration. Specifically, it includes the following steps:

[0052] Under the control of a unified time base unit, the received signals of different communication systems are resampled and the carrier frequency offset is compensated to construct a four-dimensional joint observation vector set. Specifically, this includes: driving the receiving links of each communication system with the reference clock output by the master oscillator; resampling the received signals of each communication system to a common time grid; applying the corresponding carrier frequency offset compensation amount to the resampled signals; and forming a four-dimensional joint observation vector from the four compensated signals.

[0053] Achieving strict alignment of the four received signals in both time and phase dimensions provides a physically consistent basic data structure for subsequent exploration of the intrinsic correlation between responses of different communication systems.

[0054] Extract the local coherence structure of the four-dimensional joint observation vector along the time axis, calculate the principal coherence direction within each time window, and form a time-varying joint feature trajectory. Specifically, this includes: calculating the sliding energy ratio of the received signal of each communication system, identifying energy jump points, generating an adaptive length time window centered on the energy jump point; calculating the principal coherence direction for the joint observation vector matrix within each time window, and stacking all principal coherence directions in chronological order to form a joint feature trajectory.

[0055] It depicts the co-evolution law of different communication system responses under channel mutation events, enabling the intrinsic consistency of four-dimensional signals to be stably extracted and continuously represented within a local time window.

[0056] A segmented fitting of the joint feature trajectory is performed on a four-dimensional complex unit sphere to generate a segmented constant joint channel state identifier sequence. Specifically, this includes: calculating the spherical angle between three adjacent points of the joint feature trajectory, identifying inflection points, dividing the trajectory into segments with the inflection points as boundaries, and calculating the centroid direction of each segment; verifying whether the average spherical distance from each segment point to the centroid direction meets a preset threshold; and mapping each sampling point to the corresponding trajectory segment index to generate a joint channel state identifier sequence.

[0057] By compressing the continuously changing channel response into a finite number of segmented constant state identifiers, the key nodes of channel structure transitions are preserved while the representation dimension and storage overhead are significantly reduced.

[0058] Based on the joint channel state identifier sequence, the resource allocation weights of different communication systems in each scheduling period are calculated, and a time-frequency resource allocation mapping is generated. Specifically, this includes: calculating the four-dimensional modulus distribution vector corresponding to each joint channel state identifier, defining the resource allocation weight vector based on the modulus distribution vector; counting the joint channel state identifier that appears most frequently in each scheduling period, mapping the resource allocation weights to the time-frequency resource block allocation ratio, and normalizing the total amount.

[0059] This allows resource allocation to directly reflect the actual adaptability of the current channel to various communication systems, avoiding resource misallocation and efficiency loss caused by relying on experience or fixed strategies.

[0060] Based on the time-frequency resource allocation mapping, the physical layer parameter configuration combination of each communication system within the scheduling cycle is determined to form a joint operating state tuple; specifically, this includes: predefining a finite number of physical layer operating states for each communication system, calibrating the unit resource block information carrying capacity and average maintenance time of each operating state; selecting operating states that meet the conditions based on the number of allocated resource blocks and the length of the scheduling cycle; and combining the four selected operating states into a joint operating state tuple.

[0061] Ensure that the physical layer parameters of each communication system are coordinated and adapted under channel state constraints, taking into account both information carrying efficiency and operational stability, and avoid system-level imbalance caused by aggressive configuration of a single system.

[0062] Based on the joint operating state tuple, the local oscillator phase trajectory and sampling clock phase offset of each receiving link are synchronously adjusted; specifically, this includes: storing the corresponding local oscillator phase initial value vector for each joint operating state tuple; storing the corresponding sampling clock phase offset vector for each joint operating state tuple; writing the corresponding phase value to each receiving link before the start of the scheduling period, and defining the ideal sampling time set of each receiving link within the scheduling period.

[0063] This enables the four received signals to establish a deterministic timing relationship at the analog front end, providing a reproducible and predictable physical basis for cross-system interference modeling and cancellation.

[0064] Using the sampled values ​​of each receiving link at the alignment time within the same scheduling period, a cross-system interference cancellation reference signal is constructed and injected into the corresponding receiving path; specifically, this includes: extracting the sampled values ​​of four receiving signals at the alignment time within the same scheduling period; performing sliding window processing on the sampled value sequence and calculating the projection matrix within the window; calculating the interference transfer matrix based on the projection matrix; generating a cancellation signal based on the interference transfer matrix and injecting it into the receiving path.

[0065] The analog domain closed-loop suppression is completed before the interference enters the digital domain, effectively mitigating the front-end saturation and noise rise effect of strong signal systems on weak signal systems.

[0066] After interference cancellation, the received signals from each channel are fed into a shared joint decoding unit. Based on the joint channel state identifier, the demodulated soft information from each channel is weighted and fused, and a symbol-level decision is performed. Specifically, this includes: each communication system outputs symbol-level soft information; a confidence vector is obtained based on the joint channel state identifier; each soft information is weighted according to the confidence vector; the weighted soft information is aligned by bit position; and a symbol-level fusion decision is performed. This ensures that the final decision result naturally inherits the reliability prior of the joint channel state, improving the robustness and error performance of multi-system collaborative decoding.

[0067] On the other hand, this invention proposes a system for a multi-mode fusion wireless communication chip, such as... Figure 2 As shown, it includes a unified time base unit, a signal processing unit, and a joint decoding unit, wherein:

[0068] A unified time base unit controls the synchronous sampling of signals from different communication systems across all receiving links;

[0069] The signal processing unit is used to construct a four-dimensional joint observation vector set, extract local coherent structures to form joint feature trajectories, and generate a joint channel state identifier sequence by segmental fitting on a four-dimensional complex unit sphere.

[0070] The signal processing unit is also used to calculate resource allocation weights based on the joint channel state identifier sequence, generate time-frequency resource allocation mappings, and determine the physical layer parameter configuration combinations for each communication system.

[0071] The signal processing unit is also used to adjust the local oscillator phase and sampling clock phase of each receiving link according to the physical layer parameter configuration, and to construct and inject cross-system interference cancellation reference signal;

[0072] The joint decoding unit receives the received signals after interference cancellation, weights and fuses the demodulated soft information based on the joint channel state identifier, and performs symbol-level decision-making.

[0073] Furthermore, the aforementioned unified time base unit, signal processing unit, and joint decoding unit, when executed, are also used to implement other steps of the method for a multi-mode fusion wireless communication chip, as follows:

[0074] Step 1: Construct a joint observation vector set based on the common source of multi-system received signals;

[0075] This step is based on the fundamental fact that all communication systems receive signals originating from the same transmitter and propagating through the same dynamic channel. It does not introduce any prior assumptions or external calibrations, but only obtains the original baseband samples of each system within the same time window through local synchronization mechanisms. Since there are inherent differences in the symbol period, sampling rate, center frequency, and bandwidth configuration of each system, directly splicing the sampled values ​​will lead to dimensionality mismatch and phase incomparability. Therefore, scale normalization and reference alignment must be completed first.

[0076] Step 1.1: Set up a unified time base unit inside the chip, using the 122.88MHz reference clock output from the main oscillator as the source, to drive the NB-IoT receiving link to operate at a sampling rate of 1.92MHz, the OFDM receiving link at a sampling rate of 30.72MHz, the FHSS receiving link at a sampling rate of 8MHz, and the UWB receiving link at a sampling rate of 500MHz respectively; the startup time of each link is controlled by the same trigger signal to ensure that all receiving paths are in operation. to Simultaneous activation within the time interval, including The value is 2 milliseconds, covering at least one complete NB-IoT subframe, two OFDM frames, eight FHSS frequency hopping cycles, and five hundred UWB pulse cycles.

[0077] Step 1.2: For each system... Discrete time series acquired within the window are subjected to zero-filling and resampling respectively, so that they are uniformly mapped to a common time grid. ,in Nanoseconds correspond to the highest time resolution of UWB systems; the length of NB-IoT sequences is extended after interpolation. OFDM sequence expansion to FHSS extended to UWB is natively The resampling process does not introduce additional filters; it only uses linear interpolation to maintain phase continuity and avoids introducing nonlinear distortion.

[0078] Step 1.3: Denote the four resampled sequences as follows: ;Calculate the carrier frequency offset compensation for each sequence separately for its corresponding system:

[0079] ,

[0080] ,

[0081] ,

[0082] ;

[0083] The compensation method is to multiply by a complex exponential factor. The result is recorded as This operation restores the four signals to a near-zero intermediate frequency state on the common time axis, eliminating the phase rotation dominance effect caused by the difference in center frequency between systems.

[0084] Step 1.4: Define the joint observation vector Its i-th component is ,Right now

[0085] ;

[0086] Here z[k] fully preserves the synchronous response snapshots of the four systems at the same microsecond time granularity; each dimension corresponds to the baseband complex envelope of one system, and they are not weighted or filtered to ensure the integrity of the original observations; this vector set The basic data structure that forms the basis for all subsequent processing is no longer split into individual independent sequences, nor is it subjected to dimensionality reduction preprocessing, thus providing an unmodified input source for subsequent mining of its inherent correlations.

[0087] Step 2: Extract the local coherence structure of the four-dimensional observation vector along the time axis and construct the time-varying joint feature trajectory;

[0088] Although the z[k] output in step one has a unified form, its four-dimensional components still exhibit strong non-stationarity: drastic amplitude fluctuations within a short time window, frequent phase entanglement, significant signal-to-noise ratio differences (e.g., UWB pulse peak values ​​are much higher than NB-IoT continuous wave background), and a dynamic range spanning more than three orders of magnitude. Directly performing statistical modeling on the entire K-point segment would obscure the details of local channel abrupt changes; segmenting by fixed length would make it difficult to adapt to the temporal coherence characteristics of different systems. Therefore, it is necessary to adaptively divide the analysis window based on the actual response stability of each system under the current environment, and characterize the spatial configuration evolution of the four-dimensional vector within each window.

[0089] Step 2.1: For each dimension of z[k] Calculate their sliding energy ratios respectively:

[0090] ;

[0091] in, This ratio reflects the degree of deviation of the energy at the current sampling point from the weighted average of the previous two points. The system is determined to have undergone an energy jump at time n, which may correspond to multipath arrival, occlusion initiation, or reflector reversal events; all (n,i) combinations that satisfy the conditions are recorded to form an initial mutation marker set. .

[0092] Step 2.2: Sort all markers in M ​​in ascending order of time coordinate n, merge markers with adjacent time differences less than 500 nanoseconds, and obtain the refined sequence of abrupt change moments. ; with each Extending forward and backward from the center nanosecond, in which For Let z[k] be the Frobenius norm standard deviation of a four-dimensional vector z[k] within a 1000 nanosecond window with its midpoint; thus, generate p time windows of adaptive length. Each window contains One sampling point, .

[0093] Step 2.3: For each window Extract all z[k] to form a matrix Perform column centering on the matrix, i.e., subtract the mean of each row, to obtain... Then calculate its right singular vector moments. ,satisfy ;Pick The first column As the principal coherence direction within this window, its physical meaning is: in Within a given timeframe, the four systems exhibited the most consistent pattern of change in response; For a unit-modulus complex vector, the elements This represents the participation weight and relative phase of the i-th system in this mode.

[0094] Step 2.4: Convert all p principal coherence directions Stacked in chronological order, forming joint feature trajectories. Each column of the trajectory represents a stable four-system response coordination pattern after a sudden change in channel structure. Its dimension is always 4, and its length p is determined by the dynamics of the environment. No longer carrying the original amplitude information, it only retains the pointing relationship of the four modes response on the unit sphere, thus having natural robustness to transmit power fluctuations, receive gain drift and background noise level changes; this trajectory becomes a bridge connecting transient observations and long-term channel evolution, providing a geometric basis for constructing low-dimensional representations in step three.

[0095] Step 3: Perform geodesic piecewise fitting on the joint feature trajectory on the four-dimensional complex unit sphere to generate a piecewise constant joint channel state identifier sequence;

[0096] Step 2 output It is an embedded in Medium unit sphere The discrete curves on the vector (with a magnitude of 1 and 7 degrees of freedom due to the complex four-dimensional vector) reflect the continuous evolution of the channel structure over time. However, the real channel is in an approximately steady state for most of the time, with step reconstruction occurring only at a few critical nodes. The actual shape should be composed of several approximate geodesic arc segments, each corresponding to a locally stable joint channel state. Forcibly fitting the entire trajectory with a high-order polynomial will introduce spurious oscillations; using fixed-length segments will fail to match the true radii of curvature of different arc segments. Therefore, it is necessary to automatically identify inflection points and implement a piecewise constant approximation based on the trajectory's own curvature.

[0097] Step 3.1: For three adjacent columns Calculate the cosine of the included angle between the two spheres:

[0098] ;when At that time, the judgment This is a potential inflection point; the criterion originates from the condition that the sum of the interior angles of a spherical triangle formed by three points on a unit sphere is close to π. The smaller, the more it indicates The more drastic the trajectory curvature, the more severe the inflection point; collect all j that satisfy the conditions to form an inflection point index set. .

[0099] Step 3.2: Using each of J As a boundary, Divide into m+1 segments:

[0100] Each paragraph It must contain at least 12 points; if less, merge them with the end of the previous paragraph. For each paragraph... Calculate the direction of its centroid. ,in This is the index corresponding to this segment. Its length; It is located in Above, and is the Riemann mean approximation for all points in that segment.

[0101] Step 3.3: For each segment Calculate its point to Mean spherical distance:

[0102] ;when When the radius is approximately 4.9 degrees, the segment is considered acceptable. Good representation; otherwise, further subdivide the segment until all subsegments satisfy the condition. Finally, we obtain q segments that meet the required precision, corresponding to q centroid directions. .

[0103] Step 3.4: Define the joint channel state identifier sequence ,in The assignment rule is: for each Find the time window to which z[k] belongs. Search again Let the segment index r to which it belongs be... Therefore, the entire The time window is divided into q joint channel state regions, each region consisting of a four-dimensional unit complex vector. A unique identifier; this identifier does not depend on any pilot symbols, does not involve channel impulse response estimation, and is derived solely from the inherent consistency of the received signal; its total number q is usually much smaller than K, achieving compression from millions of sampling points to dozens of state identifiers, reducing overhead by up to four orders of magnitude; this sequence S becomes the direct basis for scheduling decisions in step four.

[0104] Step 4: Generate a time-frequency resource allocation mapping relationship for multi-system cooperation based on the joint channel state identifier sequence;

[0105] Step 3 output It is a one-dimensional discrete state sequence, each Values ​​are taken from a finite set Characterized in The four time-based systems share the same joint channel configuration category. However, different systems exhibit fundamentally different sensitivities to various configurations: for example, when When the magnitude of the fourth dimension (UWB) component is close to 1 while the magnitudes of the other three dimensions are close to 0, it indicates that the current channel is highly favorable for short pulse propagation but severely suppresses continuous waves. In this case, UWB services should be prioritized. Conversely, when... When the modulus values ​​of each dimension are nearly equal and the phase difference is random, it reflects rich scattering and dense multipath propagation. OFDM systems have natural diversity gain, and their resource allocation should be increased. Therefore, a deterministic mapping from state identifiers to resource allocation needs to be established, and this mapping must meet three hard constraints: real-time execution, on-chip storage, and zero switching latency.

[0106] Step 4.1: For each status flag }, pre-calculate its four-dimensional modulus distribution vector offline. This vector reflects the relative concentration of the response energy of the four modes under this state. The closer the value is to 1, the more dominant the mode is under this configuration. Each component is The modulus of the corresponding component, and Since it is a unit vector, therefore:

[0107] This equation constitutes the basic constraint for subsequent weight allocation.

[0108] Step 4.2: Define the resource allocation weight vector:

[0109] ;

[0110] in, This design ensures And at least one component is 1, and the remaining components are scaled according to the energy percentage; Completely by The decision can be made without online calculations; simply look up a table.

[0111] Step 4.3: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require The time window is divided into N scheduling cycles, each cycle having a length of... microseconds, for For each cycle Determine the set of time points it covers. A total of 5000 sampling points were collected; the frequency of each state identifier within this set was counted to obtain a histogram. Take the identifier that appears most frequently. If there are ties, choose the one with the smallest s; This refers to the dominant joint channel state in the nth cycle.

[0112] Step 4.4: For each scheduling period n, based on Find the corresponding weight This is mapped to a time-frequency resource allocation ratio; assuming the total number of available time-frequency resource blocks in the system is R=128, where the time domain is divided into 8 orthogonal time slots and the frequency domain into 16 orthogonal sub-bands, and each resource block is the intersection of one time slot and one sub-band; let the number of resource blocks obtained by the i-th system be... ;because (Since at least one is 1, and the rest are non-negative), and Therefore The sum fluctuates between R and 4R; to ensure total conservation, a normalization adjustment is performed: Calculate ,like Then for all Then, the result is rounded down and finally fine-tuned until the sum is exactly R; this process is completed in a pipelined manner in a dedicated hardware unit within the chip, from... Input to The output time does not exceed 200 nanoseconds; the resulting The sequence is the direct execution instruction that supports reliable transmission scheduling in a multi-system collaborative manner. Its generation does not rely on pilot demodulation, query historical databases, or initiate iterative optimization. It is based solely on the joint representation structure established in steps one to three, achieving true low overhead, fast response, and strong collaboration.

[0113] Step 5: Based on the number of resource blocks allocated to each system in each scheduling cycle, determine its physical layer parameter configuration combination and solidify it into a switchable set of running states;

[0114] Step four outputs a resource block allocation sequence with a cycle of 100 microseconds. ,in This represents the total number of time-frequency resource blocks allocated to the i-th system within the n-th cycle. However, resource blocks are merely abstract containers; their actual carrying capacity depends on the specific values ​​of the physical layer parameters. For example, in NB-IoT, if single-tone mode is used, each resource block can transmit 4 bits; if multi-tone mode is used, the same resource block can transmit 16 bits, but the Doppler immunity decreases. In OFDM, shortening the cyclic prefix improves throughput but makes it more sensitive to latency spread. In UWB, narrowing the pulse width improves distance resolution but leads to a deterioration in the receiver signal-to-noise ratio. Therefore, it is necessary to... This integer mapping is a set of mutually coordinated and indivisible physical layer parameters, and this set of parameters must be able to switch at the nanosecond level without introducing phase interruption or timing misalignment.

[0115] Step 5.1: For each system i, predefine a finite set of physical layer operating states:

[0116] ,in (NB-IoT: corresponding to different combinations of subcarrier intervals and retransmission times) (OFDM: corresponding to different FFT points, cyclic prefix length and modulation order) (FHSS: corresponding to different hopping rates, spreading factors and chip rates) (UWB: corresponds to different pulse repetition frequencies, receiver integration window widths, and center frequency offsets). Each operating state Includes a complete set of parameters, for example Furthermore, all parameters are fixed via on-chip register addresses, and switching involves writing them to the corresponding address group.

[0117] Step 5.2: For each running state Offline calibration of the effective information capacity that a unit resource block can carry. (Unit: bits / block) and average duration under typical channel change scenarios (Unit: microseconds); It is determined by the modulation and coding scheme and bandwidth efficiency. The tolerance thresholds for Doppler frequency shift, time delay spread, and polarization mismatch of this state are jointly determined; both are positive real numbers, independent of environmental changes, and only related to the properties of the operating state itself; for example...

[0118] ;

[0119] and .

[0120] Step 5.3: For each scheduling period n and each system i, from Select running state To satisfy two conditions: First, ,in The average of system i across all operating states R=128 represents the total number of resource blocks; this condition ensures that the information carrying capacity of the selected state is not lower than the baseline expectation of the currently allocated resources; secondly, Microseconds, meaning the average duration of this state is no less than the length of a scheduling cycle, to avoid mid-cycle switching; when multiple When both conditions are met, the first choice is selected. The largest one, to enhance stability.

[0121] Step 5.4: Denote the selected operating state of each of the four systems within the nth cycle as... The joint running state tuples that constitute this cycle are:

[0122] All possible Forming a finite set:

[0123] The chip internally has a dedicated state index register group, each Corresponding unique index Writing to this index synchronously triggers the loading of parameters for all four receiving links; thus, The sequence was completely transformed The sequence, the latter being the simplest executable instruction form at the physical layer, no longer contains any numerical calculations or conditional judgments, only requiring table lookups and parallel writing; this conversion process is output in step four. The process is completed within 300 nanoseconds, leaving sufficient margin for cross-system signal collaborative processing in step six.

[0124] Step Six: At the beginning of each scheduling cycle, based on the joint running state tuple Synchronously adjust the phase trajectory of the local oscillator and the phase offset of the sampling clock for the four receiving links;

[0125] Step 5 output These are instructions at the scheduling level, but for the four systems to truly achieve coordinated response, they must all make consistent time calibrations for the same channel event at the same physical moment. In reality, each link oscillator has inherent frequency offset, aging drift, and temperature disturbances, and the sampling clock also experiences phase jitter due to wiring delays and power supply noise; digital domain alignment alone cannot eliminate the sub-nanosecond uncertainty introduced by the analog front end. Therefore, before the start of each cycle, according to... The indicated operating state combination implements a joint phase preset for the four local oscillators and the sampling clock, so that they form a deterministic relative relationship within the period.

[0126] Step 6.1: For each joint running state tuple Pre-store the corresponding four local oscillator phase initial value vectors. The unit is radians; this vector is not arbitrarily set, but rather determined according to the specific system. The center frequency used below With symbol period The derivation leads to: Let ,in The absolute start time of the cycle (with the main time base as a reference). This is a fixed phase compensation amount for the system in the corresponding operating state, used to compensate for known circuit path delays; The chip is calibrated once and fixed in OTP before leaving the factory, and it does not change with the environment.

[0127] Step 6.2: For each Simultaneously, it stores the phase offset vectors of the four sampling clocks relative to the main time base. The unit is radians; this offset ensures that the first sampling point of each link within the period falls near the optimal decision point of its respective symbol waveform; its value is determined based on:

[0128] ,in Let i be the sampling rate of the i-th path. For this system in The symbol period below, This is the sum of the receiver filter group delay and the ADC aperture delay of this system, in seconds; It is a constant and has been calibrated through on-chip loopback testing.

[0129] Step 6.3: 500 nanoseconds before the start of the nth cycle, the chip control unit reads... Obtained by looking up the table The initial value is then written to the phase accumulator of the four-channel local oscillator phase-locked loop. Write to the phase bias register of the four-channel sampling clock divider All write operations are triggered on the same master clock edge, ensuring strict synchronization of the four channels; after writing, each local oscillator operates at its own frequency. It operates freely, but the initial phase has been precisely preset; the sampling clocks of each channel are also set accordingly. Free counting, but the first sampling edge time has been offset. Second.

[0130] Step 6.4: Define the ideal sampling time set of the four received signals within period n as:

[0131] ,in The absolute moment of the start of the period. Let i be the number of samples that should be collected in the i-th path during this period. Microseconds are the period length; because It has been precisely set, therefore All sampling times are aligned with the symbol structure of each system, with no interpolation or missing points. More importantly, the sampling times of different systems exhibit a regular, staggered rather than random distribution on the time axis. For example, when the OFDM symbol boundary falls precisely near the peak of the UWB pulse, its sampling point will naturally capture the strongest energy component. This... The dominant temporal coordination relationship becomes the physical basis for cross-system interference suppression in step seven.

[0132] Step 7: Using the joint sampled values ​​of the four received signals at the alignment time within the same scheduling period, construct a cross-system interference cancellation reference signal and inject it into the receiving path in real time;

[0133] Step six ensures a deterministic temporal correlation between the four samples, but inherent coupling still exists between the different systems: for example, strong UWB pulses can cause broadband saturation at the OFDM receiver front end, leading to distortion of dozens of subsequent OFDM symbols; after the NB-IoT continuous wave signal enters the UWB receiving channel, harmonic components remain even after bandpass filtering, raising the noise floor; if the FHSS frequency hopping signal momentarily overlaps with the OFDM subcarrier, it will cause burst errors. Traditional solutions rely on independent filtering or power back-off, sacrificing efficiency and sensitivity. This step instead utilizes the four aligned sample values, using the interference component observed by one system as a cancellation reference for another system, achieving on-chip closed-loop suppression.

[0134] Step 7.1: For each scheduling cycle n, extract the four paths in The sampled values ​​at the corresponding time points are used to form a four-dimensional vector sequence. ,in , The i-th component is The most direct The sampled values ​​( (nanoseconds), padding with zeros if necessary; the sequence length is uniformly L, preserving the original time alignment.

[0135] Step 7.2: For Perform sliding window processing along the k-dimensional plane, with a window length W=64 and a step size of 1; for each window... Calculate its column space projection matrix. The matrix projects any vector onto a subspace spanned by the four responses of the current window. If a certain system i is mainly an interference source within this window (e.g., the UWB pulse energy is much higher than the other three), then its component will dominate the result after projection.

[0136] Step 7.3: Define the interference transfer matrix Its element in the i-th row and m-th column is ;

[0137] in The i-th component, i.e., the sampled value of the i-th system at time k; the physical meaning of this formula is: within window j, the covariance contribution of the original sample of the m-th system to the projection result of the i-th system; when When the value is large, it indicates that there is a significant linear interference mapping between UWB (path 4) and NB-IoT (path 1); It is a real symmetric matrix that is updated once per period and does not store history.

[0138] Step 7.4: Receive the original received signal of the i-th mode within the n-th period. At each sampling point Generate cancellation signal Where j is the window index covering t; after digital-to-analog conversion, the cancellation signal is injected as a current into the input of the low-noise amplifier of the i-th receiver link, with the polarity opposite to the original interference; the injection strength is determined by... Directly determined, no additional gain adjustment required; due to Based on real-time sampling calculation and with the injection point located at the analog front end, strong interference that has not yet entered the digital domain can be suppressed, and the timing coordination established in step six can be transformed into an actual signal-to-interference ratio improvement. This process is completed within the cycle without occupying an additional scheduling cycle or changing the original demodulation process.

[0139] Step 8: Send the four received signals after interference cancellation within each scheduling cycle to the shared joint decoding unit, based on the joint channel state identifier. Execute a symbol-level confidence-weighted fusion decision;

[0140] Step seven outputs four clean sampled sequences after analog domain interference suppression, but each system is still demodulated independently, failing to utilize the joint state identifier. The implied prior knowledge of the channel structure. For example, when correspond When the magnitudes of OFDM and UWB components are similar and their phases are approximately orthogonal, it indicates that the channel possesses both good multipath diversity and short-time impulse response characteristics. In this case, the soft bits demodulated by OFDM and the arrival timestamps demodulated by UWB can be used for mutual verification. If the magnitudes of both NB-IoT and FHSS components are low, it indicates the presence of strong narrowband blocking, and their demodulation results should be downweighted overall. Therefore, a shared decoding unit needs to be established to fuse the four demodulated outputs at the symbol level, and based on... Dynamically adjust the confidence weights of each path.

[0141] Step 8.1: For each scheduling cycle n, the four receiving links complete the standard demodulation of their respective systems and output symbol-level soft information: NB-IoT output. ( (Number of demodulated symbols, 2 bits per symbol) OFDM output (4 bits per symbol), FHSS output (1 bit per symbol), UWB output (Each symbol is 1 bit, representing the time of arrival quantization value); all soft information is represented in log-likelihood ratio (LLR) form, with positive values ​​indicating "0" being more reliable and negative values ​​indicating "1" being more reliable.

[0142] Step 8.2: Based on the results obtained in Step 3 The corresponding joint channel state confidence vector is obtained by looking up the table. ,in This value directly reflects the proportion of response energy of the i-th system in this state, that is, the inherent reliability of its demodulation result under the current channel configuration. For example, if correspond .

[0143] Step 8.3: Weight the four soft information streams separately.

[0144] After weighting, the LLR amplitude of each path strictly corresponds to its channel adaptability; then, all weighted soft information is aligned according to bit position: for each service stream (such as sensor reporting frames), the common information bits that may be carried in the four paths are extracted to form a joint LLR vector. ,in This is the index of the bit in the i-th path; if a path does not carry the bit, the corresponding position is filled with 0.

[0145] Step 8.4: For each joint LLR vector Perform symbol-level decision: The final bit estimate is:

[0146] ;

[0147] in for The i-th component; this decision does not introduce new parameters, but only performs algebraic summation, requiring only an adder and a comparator in hardware implementation; all decision results are reassembled according to the traffic flow to form a unified output frame; this frame is neither a simple output of a particular system nor a simple majority vote, but rather a result of joint channel states. The weighted fusion result guided throughout the process has a lower error probability than any single-mode independent decision; thus, the complete closed loop of the multi-mode fusion wireless communication chip method and system is completed, supporting reliable transmission scheduling for multi-mode collaboration.

[0148] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for using a multi-mode fusion wireless communication chip, characterized in that, include: Under the control of a unified time base unit, resampling and carrier frequency offset compensation are performed on the received signals of different communication systems to construct a four-dimensional joint observation vector set; Extract the local coherence structure of the four-dimensional joint observation vector along the time axis, calculate the principal coherence direction within each time window, and form a time-varying joint feature trajectory; Segmented fitting is performed on the joint feature trajectory on a four-dimensional complex unit sphere to generate a segmented constant joint channel state identifier sequence. Based on the joint channel state identifier sequence, the resource allocation weights of different communication systems in each scheduling period are calculated to generate a time-frequency resource allocation mapping. Based on the time-frequency resource allocation mapping, the physical layer parameter configuration combination of each communication system within the scheduling cycle is determined to form a joint operating state tuple. Based on the joint operating state tuple, the local oscillator phase trajectory and sampling clock phase offset of each receiving link are synchronously adjusted. Using the sampled values ​​of each receiving link alignment time within the same scheduling period, a cross-system interference cancellation reference signal is constructed and injected into the corresponding receiving path. The interference-cancelled received signals are sent to the shared joint decoding unit. Based on the joint channel state identifier, the demodulation soft information of each path is weighted and fused, and symbol-level decision is performed. The construction of the four-dimensional joint observation vector set includes: driving the receiving links of each communication system with the reference clock output by the master oscillator; resampling the received signals of each communication system to a common time grid; applying the corresponding carrier frequency offset compensation to the resampled signals; and forming a four-dimensional joint observation vector from the four compensated signals. The step of determining the physical layer parameter configuration combination of each communication system within the scheduling cycle based on the time-frequency resource allocation mapping includes: predefining a finite number of physical layer operating states for each communication system, and calibrating the unit resource block information carrying capacity and average maintenance time of each operating state; selecting operating states that meet the conditions based on the number of allocated resource blocks and the length of the scheduling cycle; and combining the four selected operating states into a joint operating state tuple.

2. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The extraction of the local coherence structure of the four-dimensional joint observation vector along the time axis includes: Calculate the sliding energy ratio of the received signals of each communication system, identify the energy jump point, and generate an adaptive time window centered on the energy jump point; Calculate the principal coherence direction for the joint observation vector matrix within each time window, and stack all principal coherence directions in chronological order to form a joint feature trajectory.

3. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The segmental fitting of the joint feature trajectory on a four-dimensional complex unit sphere includes: Calculate the spherical angle between three adjacent points of the joint feature trajectory, identify the inflection point, divide the trajectory into segments with the inflection point as the boundary, and calculate the centroid direction of each segment; Verify whether the average spherical distance from each segment point to the centroid direction meets the preset threshold; Each sampling point is mapped to the corresponding trajectory segment index to generate a joint channel state identifier sequence.

4. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The generation of the time-frequency resource allocation mapping includes: Calculate the four-dimensional modulus distribution vector corresponding to each joint channel state identifier, and define the resource allocation weight vector based on the modulus distribution vector; The joint channel state identifier that appears most frequently in each scheduling cycle is statistically analyzed, and the resource allocation weights are mapped to the time-frequency resource block allocation ratios, and the total amount is normalized.

5. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The synchronous adjustment of the local oscillator phase trajectory and sampling clock phase offset of each receiving link includes: Store the corresponding local oscillator phase initial value vector for each joint running state tuple; Store the corresponding sampling clock phase offset vector for each joint running state tuple; Before the start of the scheduling period, the corresponding phase value is written to each receiving link, and the ideal sampling time set of each receiving link within the scheduling period is defined.

6. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The constructed cross-system interference cancellation reference signal includes: Extract the sampled values ​​of the four received signals at the alignment time within the same scheduling period; Apply a sliding window processing to the sampled value sequence and calculate the projection matrix within the window; The interference transfer matrix is ​​calculated based on the projection matrix, and a cancellation signal is generated and injected into the receiving path based on the interference transfer matrix.

7. The method for a multi-mode fusion wireless communication chip according to claim 1, characterized in that, The execution of symbol-level decisions includes: Each communication system outputs symbol-level soft information, and obtains a confidence vector based on the joint channel state identifier; Each soft information is weighted according to its confidence vector, and the weighted soft information is aligned according to its bit position before a symbol-level fusion decision is performed.

8. A system for implementing a multi-mode fusion wireless communication chip according to any one of claims 1-7, characterized in that, It includes a unified time base unit, a signal processing unit, and a joint decoding unit, wherein: A unified time base unit controls the synchronous sampling of signals from different communication systems across all receiving links; The signal processing unit is used to construct a four-dimensional joint observation vector set, extract local coherent structures to form joint feature trajectories, and generate a joint channel state identifier sequence by segmental fitting on a four-dimensional complex unit sphere. The signal processing unit is also used to calculate resource allocation weights based on the joint channel state identifier sequence, generate time-frequency resource allocation mappings, and determine the physical layer parameter configuration combinations for each communication system. The signal processing unit is also used to adjust the local oscillator phase and sampling clock phase of each receiving link according to the physical layer parameter configuration, and to construct and inject cross-system interference cancellation reference signal; The joint decoding unit receives the received signals after interference cancellation, weights and fuses the demodulated soft information based on the joint channel state identifier, and performs symbol-level decision-making.

Images