A dynamic key-based intercom communication method

Abstract

The application provides a kind of intercom communication method based on dynamic key, it is related to intercom communication technical field, including: when call is established, the PTT behavior parameter of communication party is collected, and behavior entropy component is obtained by quantization processing;The statistical stability of the historical carrier frequency offset sequence of each device is evaluated, and the frequency offset baseline is obtained;Through key negotiation protocol and key derivation, the session key bound with the behavior of both sides and hardware characteristics is generated;During the call, the safety margin of the session key is comprehensively evaluated, and the negotiation and caching of new session key are completed in advance in the PTT mute window;At the same time, the real-time change of carrier frequency offset is continuously monitored;When the call is over, the key gene is updated by rolling using one-way hash chain, and the current session key is cleared.The application can solve the problem that the key material is easy to be reconstructed offline in the half-duplex communication scene of intercom, and the key refresh is decoupled from real-time security threats.

Description

Technical Field

[0001] This invention relates to the field of walkie-talkie communication technology, and in particular to a walkie-talkie communication method based on dynamic keys. Background Technology

[0002] Two-way radio communication systems generally employ a half-duplex operating mode, with both parties alternating between sending and receiving via the PTT (Push-To-Talk) protocol, and only one party occupying the channel at any given time. Because sending and receiving cannot occur synchronously, the two parties cannot sample the same wireless channel simultaneously, making Physical Layer Key Generation (PLKG) technology, based on the reciprocity of wireless channel impulse response, unsuitable for this scenario. Existing secure two-way radio communication standards (such as TETRA and DMR) generally employ Over-the-Air Key Transfer (OTAR) schemes, relying on a centralized key management server to distribute session keys to terminals. Key refresh cycles are typically triggered by fixed time intervals or accumulated data transmission volume, decoupled from the actual security status during communication. Once key materials are leaked, attackers can reconstruct historical keys offline using known parameters, posing a significant risk of key material reconstruction.

[0003] Chinese patent application CN115278667A discloses a method for encrypted communication of public network walkie-talkies. This scheme constructs a three-layer key management system consisting of a walkie-talkie terminal, a Management and Dispatch Console (MDTT), and a Key Security Center (AKSC). The AKSC is responsible for generating and managing group keys, and pre-filling quantum keys into the terminal's local secure medium (SM) through a Key Interface Library (KIL). During a call, the PTT module randomly selects a group key from the local key library to encrypt the session data, and the receiving end matches and decrypts the data based on the group key index carried in the protocol packet. Furthermore, the scheme introduces an authentication process for the secure medium to verify the terminal's legitimacy. However, the group keys in this scheme are still updated uniformly at fixed time intervals, and the management of key validity is completely unrelated to the real-time physical state and communication behavior of each terminal, making it impossible to dynamically adjust the key lifecycle based on security anomalies that occur during device operation. Summary of the Invention

[0004] In view of this, the present invention provides a walkie-talkie communication method based on dynamic keys. By quantifying two types of signals that can be acquired in real time, namely PTT protocol layer behavioral characteristics and carrier frequency offset (CFO), these are used as the entropy source of key material and the physical layer protection indicator of key validity period, respectively. This binds key generation to the behavior of terminal operators and the characteristics of radio frequency hardware, and makes key revocation and refresh respond to changes in the actual physical state of the device. Thus, without relying on a centralized key distribution infrastructure, the problem of easy offline reconstruction of key material and decoupling of key refresh from real-time security threats in half-duplex walkie-talkie communication scenarios is solved.

[0005] The technical solution of this invention is implemented as follows: This invention provides a walkie-talkie communication method based on dynamic keys, comprising: When a call is established, PTT behavior parameters of both parties are collected and quantified to obtain behavior entropy components; statistical stability of the historical carrier frequency offset sequence of each device is evaluated to obtain the frequency offset baseline; using the behavior entropy components and the frequency offset baseline as key genes, a session key bound to the call behavior and hardware characteristics of both parties is generated through key negotiation protocol and key derivation. During the call, the security margin of the session key is comprehensively evaluated, and the negotiation and caching of the new session key are completed in advance within the PTT silent window. At the same time, the real-time changes of the carrier frequency offset are continuously monitored and compared with the frequency offset baseline according to the rate of change. If the rate of change is within the normal range, the session key remains valid. When an abnormal change is detected, the current session key is revoked. If the new session key is already stored in the cache, it is immediately enabled to restore communication; otherwise, the key negotiation is re-initiated. When the call ends, the key gene is updated using a one-way hash chain, and the current session key is cleared.

[0006] Preferably, the PTT behavior parameter includes three components: button duration. Delay from the moment the key is pressed to the moment the voice activity detection is triggered. The short-time energy mean of the first frame of audio signal after voice activity detection is triggered. ; Key press duration Delay with short-time energy mean The three components constitute the PTT behavior vector for this call, which is accumulated by sliding within the acquisition window of K consecutive calls to form a behavior matrix. , where K is the preset acquisition window length.

[0007] Preferably, the behavioral entropy component is derived from the permutation entropy method for the behavioral matrix. Quantization, behavioral entropy components Defined as: ; Where i is the terminal index. The permutation entropy value is calculated for the row matrix with embedding dimension m. This represents the theoretical maximum value of the permutation entropy under embedding dimension m. The normalized value range is within (0,1) for the behavior entropy component; the embedding dimension m is a preset parameter.

[0008] Preferably, the key gene is constructed through a cascaded hash operation, and the key gene of terminal i... Defined as: ; in, It is a one-way hash function. Indicates byte concatenation; This is the device's inherent identifier entropy value. This is the timestamp for this call. The channel noise sample value, This is the call sequence number. This is the output after the behavioral entropy components have been independently hashed. It is a scalar normalized to the carrier frequency offset measurement value after rounding with a fixed quantization step size.

[0009] Preferably, the key negotiation protocol adopts the elliptic curve Diffie-Hellman protocol, and the key derivation adopts a hash-based key derivation function; The calling terminal i and the called terminal j each generate a temporary private key using a cryptographically secure random number generator. And calculate the corresponding temporary public key. , , where G is the base point of the elliptic curve; After exchanging temporary public keys, both parties calculate the shared secret separately. ; Using the key genes of both parties Call timestamps serve as a source of additional entropy. Salt value, call sequence number For contextual information, derive the session key using the key derivation function: ; in It is injected as an additional entropy source and does not participate in the generation of temporary private keys.

[0010] Preferably, the rate of change comparison employs a dual-rate judgment mechanism: The preamble synchronization frame carrying the measured carrier frequency offset each time the PTT button is pressed carries the actual value. , where t is the measurement order index; The frequency offset baseline is denoted as The receiver calculates the measured value and the frequency offset baseline. Deviation amount: ; And calculate the short-time mutation rate between two adjacent measurements: ; in This is the deviation obtained from the previous measurement; when Exceeding the short-term mutation rate threshold If the event is detected as an abnormal mutation, the session key will be revoked. when Exceeding the long-term cumulative deviation threshold but Not exceeding At that time, the frequency offset baseline was adjusted using the average of nearly W measurements. Perform adaptive updates and record the cumulative baseline drift. ; when The session key remains valid during this period.

[0011] Preferably, the comprehensive evaluation uses the key strength evaluation value KS to quantify the security margin of the session key, and is calculated at each PTT release time: ; Where i is the terminal index, q is the cumulative number of encrypted packets in this session, and SNR is the measured signal-to-noise ratio of the current channel. As the initial reference signal-to-noise ratio, The channel power-law stress coefficient; For the current behavior entropy component, The mean entropy of historical behavior. This represents the deviation of the current behavior entropy component from the historical mean. This is the stress sensitivity coefficient for abnormal behavior; This represents the cumulative baseline drift. This is the cumulative stress coefficient for carrier frequency offset drift. To preset a safety lower limit threshold; when KS is lower than At that time, a new session key pre-negotiation is triggered within the silent window.

[0012] Preferably, the silence window is an inherent silence time window between the release of the PTT button and the next PTT press. There is no audio service transmission within this window, but the wireless link remains active; when At that time, the terminal is A new round of key negotiation is completed within the current call, and the new session key is cached locally. The current session key remains valid until the end of the current call, and the cached new session key is used directly when the next call is established. During this period, pre-negotiation is not performed within the silent window, and the current session key continues to be used.

[0013] Preferably, in a group call scenario, the group coordinator collects the key data reported by each of the n member terminals in the group, and splits the group session key into n shared shares using a threshold secret sharing scheme. A reconstruction threshold of 'a' is set, and each share is encrypted and distributed to the corresponding terminal. Shares from any 'a' terminals can be used to reconstruct the group session key. Any member terminal within the group can determine locally that KS is lower than 'a'. When a request for a group key refresh is sent to the group coordinator, the group coordinator initiates a new round of group key synthesis and distribution within the silent window of the current call round; the KS obtained by the member terminal with the lowest key strength evaluation value is used as the basis for determining the group key refresh.

[0014] Preferably, at the end of the call, a one-way hash chain is performed to update the key gene. The terminal uses the current key gene as input and calculates the next key gene using a one-way hash function. ; Where i is the terminal index. The key gene used for the k-th call. This is the call sequence number. It is a one-way hash function; after the update is completed, the session key is immediately cleared in memory.

[0015] The present invention has the following advantages over the prior art: (1) This method uses the permutation entropy quantization value of PTT behavior parameters and the carrier frequency offset baseline to form a key gene, and injects it as an additional entropy source into the derivation process of the session key. Since the above two types of parameters are associated with specific operators and specific device hardware, it is difficult to reconstruct equivalent key materials in an offline environment without the corresponding terminal and operation records.

[0016] (2) This method continuously monitors the short-term change rate and long-term cumulative deviation of the carrier frequency offset to determine whether the terminal's radio frequency hardware status is abnormal, and triggers key revocation or baseline update accordingly. Compared with the method of refreshing the key only at fixed time periods, this mechanism makes the key validity period management correspond to the actual operating status of the terminal, and can respond in a timely manner when hardware status abnormalities are detected.

[0017] (3) This method completes the pre-negotiation and caching of the new key within the PTT silent window, without requiring a separate voice transmission period for key refresh. For half-duplex communication scenarios, this method minimizes the impact of the key update process on the call flow.

[0018] (4) After the call ends, this method updates the key gene using a one-way hash chain by introducing the average entropy of the current session, and clears the session key in memory. Due to the one-way nature of the hash function, the current key gene cannot be used to derive historical session keys, thus enabling completed historical sessions to have forward security. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the PTT behavior entropy quantification process of the present invention; Figure 3 This is a flowchart of the CFO baseline establishment and degradation process of the present invention; Figure 4 This is a schematic diagram of the key gene and key derivation of the present invention; Figure 5 This is a schematic diagram of the CFO dual-rate protection and key refresh mechanism of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, the present invention provides a walkie-talkie communication method based on dynamic keys, comprising: When a call is established, the PTT (Push-To-Talk) behavior parameters of both parties are collected and quantified to obtain the behavior entropy component; the historical carrier frequency offset sequence of each device is statistically stable to obtain the frequency offset baseline; using the behavior entropy component and the frequency offset baseline as the key gene, a session key bound to the call behavior and hardware characteristics of both parties is generated through key negotiation protocol and key derivation. During the call, a comprehensive assessment of the security margin of the session key is conducted, and the negotiation and caching of the new session key are completed in advance within the PTT silent window. At the same time, the real-time changes of the carrier frequency offset (CFO) are continuously monitored and compared with the frequency offset baseline according to the rate of change. If the rate of change is within the normal range, the session key remains valid. If an abnormal change is detected, the current session key is revoked. If a new session key is already stored in the cache, it is immediately enabled to restore communication; otherwise, key negotiation is re-initiated. When the call ends, the key gene is updated using a one-way hash chain, and the current session key is cleared.

[0023] This invention executes according to three natural stages of the key lifecycle: In the call setup stage, behavioral entropy quantification, CFO baseline establishment, key gene construction, and session key negotiation and derivation are completed; in the call execution stage, two parallel mechanisms—the CFO dual-rate judgment mechanism and the key strength comprehensive evaluation—jointly safeguard the session key. Group call scenarios, as an extension of point-to-point calls, are described separately in the call execution stage; in the call termination stage, the one-way hash chain update of the key gene and the session key clearing are performed. The CFO dual-rate judgment and KS (Key Strength) evaluation are continuously performed in parallel during the call, calculated independently at the PTT button trigger and PTT release time, respectively, without any execution dependency between them.

[0024] like Figure 2 As shown, at the start of each call, the terminal collects three PTT behavior components: button press duration. Defined as the duration from pressing to releasing a physical button in a PTT (Personal Tone Telephone) system; the delay from the moment the button is pressed to the moment voice activity detection is triggered. Defined as the time difference between the moment a key event is detected by the protocol layer and the moment the VAD module (Voice Activity Detection) outputs the first valid voice frame flag; the short-time energy average of the first frame audio signal after voice activity detection is triggered. , is defined as the average energy of the sampling points of the first frame of audio after VAD is triggered.

[0025] The above three components The PTT behavior vector constituting this call is accumulated by sliding within the acquisition window of K consecutive calls to form a behavior matrix. , where K is the preset acquisition window length. for The matrix is ​​3D, where each row corresponds to a behavior vector of a call and each column corresponds to three behavior components. When a new call is added, the earliest record is discarded, and the window length is kept constant at K.

[0026] Behavior Matrix After construction, the permutation entropy method is used to quantify it, obtaining the normalized behavioral entropy components. Defined as: ; Where i is the terminal index. The permutation entropy value is calculated for the row matrix with embedding dimension m. This represents the theoretical maximum value of the permutation entropy under embedding dimension m. The behavior entropy component whose normalized value range is (0,1) is defined as the embedding dimension m, which is a preset parameter. The specific calculation process is as follows: For... For each column, extract a sliding subsequence of length m according to the embedding dimension m, and map each subsequence according to the element size relationship. One of the permutation patterns is to count the frequency of occurrence of each permutation pattern and calculate the Shannon entropy. ,in Let be the frequency of the j-th permutation pattern. The average of the three calculated results is then used to... Normalization yields . A value close to 1 indicates that the operator's PTT behavior is highly random within the window, while a value close to 0 indicates that the behavior is highly regular.

[0027] like Figure 3 As shown, while performing behavioral entropy quantization, the terminal performs a statistical stability assessment on its historical carrier frequency offset sequence to establish a frequency offset baseline. Let the historical CFO measurement sequence accumulated by terminal i before establishing the call be . Calculate the mean of the sequence. and standard deviation The coefficient of variation is defined as .when When a preset stability threshold is reached, the CFO sequence is deemed to have sufficient stability, based on the sequence mean. As frequency offset baseline .when If the current CFO sequence is not stable enough, the terminal will continue to participate in the key gene construction by using the valid baseline value (if it exists) retained at the end of the previous call. If the terminal is making its first call and has no historical valid baseline, the historical sequence accumulation window N needs to be extended until the stability condition is met before a session can be established. At the same time, a status flag indicating that the baseline is not ready is sent to the other end. After both parties negotiate and confirm, the key gene construction step will be entered to ensure that the source of the CFO component in the key gene is reliable.

[0028] like Figure 4 As shown, the key gene is constructed through a concatenated hash operation, and the key gene of terminal i is... Defined as: ; in, It is a one-way hash function. Indicates byte concatenation; The device's inherent identifier entropy value is obtained by entropy extraction processing from the terminal hardware identifier. This is the timestamp for the call, with an accuracy of at least seconds, used to bind the session duration; The channel noise sample value is a short-term sample taken during the channel idle state when the call is established, which is used to introduce channel randomness. This is the call sequence number, which is strictly incremented after each call to distinguish different rounds of the call. The output of the behavior entropy component after independent hashing is written into the key gene in hash form to prevent reverse calculation of the behavior benchmark. This is a scalar normalized to the carrier frequency offset measurement value after rounding with a fixed quantization step size. The selection of the quantization step size ensures that different measurement values ​​within the normal temperature drift range are mapped to the same integer interval, eliminating the influence of measurement noise while preserving individual hardware differences between devices. (Key Gene) The logic of introducing each component is complementary: Bind device identity, and To ensure that the key genes for different rounds of communication are not duplicated, Introducing channel randomness to prevent offline pre-computation and The operator's behavior and the radio frequency hardware characteristics are respectively bound.

[0029] Key gene and During call setup, the key exchange is completed via the signaling channel. Before the exchange, each terminal uses the peer's temporary public key to encrypt and encapsulate its own key gene, ensuring that the key gene is not exposed in plaintext during transmission over the signaling channel, preventing... Hardware characteristics were intercepted.

[0030] The key negotiation protocol adopts the elliptic curve Diffie-Hellman protocol, and the key derivation adopts a hash-based key derivation function; The calling terminal i and the called terminal j each generate a temporary private key using a cryptographically secure random number generator (CSPRNG). And calculate the corresponding temporary public key. Where G is the base point of the elliptic curve. The temporary private key is generated independently by CSPRNG. and It does not participate in the generation of the temporary private key to ensure that the cryptographic security of ECDH (Elliptic Curve Diffie-Hellman) key exchange is not constrained by behavioral entropy or the quality of the CFO component. After exchanging temporary public keys, both parties respectively compute the shared secret. As can be seen from the mathematical properties of elliptic curves, the calculation results on both sides are equal. (Based on the key genes of both parties...) Call timestamps serve as a source of additional entropy. Salt value, call sequence number For contextual information, derive the session key using the key derivation function: ; The key derivation function uses a hash-based key derivation function (HKDF, HMAC-based Key Derivation Function), which internally performs two phases: extraction and expansion. The extraction phase... For salt, The input key material is used to output a pseudo-random key; the expansion phase uses... The context binding information is used to output the session key SeK of the required length. The HKDF extraction phase homogenizes the input entropy. and The entropy contribution of each component is injected in a probabilistic mixing manner. Even if the information content of a certain component is low, it will not weaken the ECDH security strength provided by Z. The two are cryptographically independent. Due to the... Both injection derivation processes require attackers to simultaneously possess the hardware CFO characteristics and behavioral records of both devices to reproduce the session key; information from only one party is insufficient for offline reconstruction of the SeK.

[0031] like Figure 5 As shown, during the call, the receiving end continuously monitors the carrier frequency offset in real time. Each time the PTT button is triggered, the sending end carries the measured value of its own carrier frequency offset in the preamble synchronization frame. , where t is the measurement order index. The receiving end extracts... Then, the measured values ​​and the frequency offset baseline are calculated. Deviation amount: ; And calculate the short-time mutation rate between two adjacent measurements: ; in This represents the deviation obtained from the previous measurement.

[0032] The dual-rate determination mechanism handles the following three cases respectively: when Exceeding the short-term mutation rate threshold At that time, it was determined to be an abnormal change. The normal temperature drift of the CFO is determined by the crystal oscillator temperature coefficient, and the rate of change is usually slow. However, physical replacement of equipment or radio frequency spoofing attacks can cause it to change. There was a significant jump between two consecutive PTT button presses. This will significantly exceed the normal temperature drift rate range. Upon triggering an abnormal mutation detection, the receiving end immediately revokes the current session key; if a pre-negotiated new session key is already stored in the cache, it is immediately enabled to restore communication; otherwise, an emergency key negotiation request is initiated to the peer, and the complete ECDH negotiation process is re-executed. A revoked SeK must not be reused under any circumstances.

[0033] when Exceeding the long-term cumulative deviation threshold but Not exceeding If the baseline drift is determined to be due to normal temperature drift, key revocation will not be triggered. Instead, the frequency offset baseline will be adjusted based on the average of the last W measurements. Perform adaptive updates: Simultaneously, the step size of this baseline update is recorded and added to the cumulative baseline drift. ,in For the updated baseline value, The baseline value before the update; It serves as a cumulative measure of the long-term drift of the CFO and is used in subsequent KS calculations.

[0034] when At that time, the measured CFO was within the normal range near the baseline, so there was no need to update the baseline and the session key remained valid.

[0035] In practical deployments, the short-time mutation rate threshold The accuracy of this directly determines the reliability of the abnormal mutation judgment. If... If the setting is too low, normal temperature drift, which occurs at a reasonable rate when the ambient temperature changes rapidly, will be misinterpreted as an abnormal mutation, leading to the unnecessary revocation of the key; if Setting the threshold too high may miss actual CFO fluctuations caused by equipment replacement. Different devices have individual variations in crystal oscillator temperature coefficients, and the factory-preset fixed thresholds cannot accurately adapt to the actual temperature drift characteristics of each device. Given that the historical CFO sequence has been completed during the CFO baseline establishment phase... Based on the statistical evaluation, the first-order difference sequence of the sequence can be further calculated. The sample distribution of CFO changes between measurements during the baseline establishment period was obtained, and the standard deviation of this sample distribution was used as the basis for the calculation. As the actual measurement benchmark for the short-term change rate of the machine's CFO, it is calibrated according to the following formula. : ; Where k is a preset multiplier factor. The standard deviation of the first-order difference sequence of CFO during the baseline establishment period is given. The calibration data, derived from on-site measurements rather than factory-set parameters, naturally reflects the temperature coefficient characteristics of the individual crystal oscillator. The calibration process reuses the existing CFO history sequence from the baseline establishment step, without introducing additional measurement or storage overhead. The k-value can be pre-configured based on the device's operating environment (a smaller value for indoor fixed deployments and a larger value for outdoor high-temperature differential environments) to adapt to the diverse deployment scenarios of walkie-talkies. This calibration is performed synchronously upon baseline establishment, and the calibration results are written into the threshold parameters of the dual-rate judgment mechanism. Its lifespan is correlated with the frequency offset baseline. Maintain consistency and jointly maintain the same adaptive logic.

[0036] Key strength assessment and CFO monitoring run in parallel. At each PTT release time, the terminal calculates the key strength assessment value KS. ; Where q represents the total number of encrypted packets in this session. This term originates from birthday boundary analysis: For a 128-bit block cipher, this term decreases linearly with increasing q, reflecting the depletion of security margin due to continuous key use; SNR is the measured signal-to-noise ratio of the current channel. The initial reference signal-to-noise ratio when the call is established. The channel power-law stress coefficient reflects the compression of the security margin due to the reduced difficulty of ciphertext interception when the channel quality deteriorates. For the current behavior entropy component, The mean entropy of historical behavior. This represents the deviation of the current behavior entropy component from the historical mean. The abnormal behavior stress sensitivity coefficient reflects the contribution of abnormal deviations in operator behavior to the attenuation of key security margin. This represents the cumulative baseline drift. This is the cumulative stress coefficient for carrier frequency offset drift. This factor reflects the erosion of the credibility of the CFO component in the key gene by the long-term drift of the device's RF hardware state. To preset the lower safety threshold, when The new session key pre-negotiation within the silent window is triggered at any time. The KS formula adopts a product structure to ensure that any deterioration of any one of the four factors—ciphertext accumulation, channel quality, behavioral anomalies, and hardware drift—can cause a decrease in KS, and there is no mutual compensation or masking among the factors.

[0037] In this invention, the behavioral entropy component It simultaneously performs two functional roles: firstly, it processes data through hashing during call establishment... Write key genes in form Firstly, it participates in session key derivation, playing a role in binding operator behavior; secondly, it is used during the call. The form appears in the KS formula as a quantitative indicator for anomaly detection, affecting the key strength evaluation result. This is precisely... This dual role makes The accuracy of the (mean of historical behavior entropy) affects both the reliability of the key gene and the accuracy of the KS evaluation. When the operator's PTT behavior undergoes a slow, systematic drift due to changes in task intensity or communication scenario switching, if By using a simple moving average to track and update, drifting samples will be gradually incorporated. lead to The sensitivity to changes in actual behavior decreases, and the abnormal stress term in KS approaches 1 and no longer plays a substantial role; at the same time, if the drifted Continuous writing Then, the behavioral binding component in the key gene gradually deviates from the original operator's behavioral characteristics, reducing the ability to identify situations where the device is being used by others. To address this issue, at the end of each call, the current... and The deviation amount, and with Historical variance estimates Comparison: When ( Normal update when the preset tolerance factor is used. : ; in The forgetting factor takes values ​​of ,make Assign higher weights to recent behavior and exponentially decaying weights to long-term behavior; when At that time, freeze The update will not change the current The call is included in the mean calculation and simultaneously marked as an abnormal behavior event, triggering a reassessment of KS. This mechanism will... Decoupling update rights from anomaly detection rights: during the evolution of normal behavior By smoothing the tracking through the forgetting factor, the KS is guaranteed. The validity of the baseline; when behavior changes abruptly Freezing ensures that anomalies are not absorbed by the mean, allowing KS to accurately detect and trigger key pre-negotiation. This processing logic directly affects the KS formula. The calculation of the item, and the key gene The build process does not involve interface changes, and the two are independent of each other.

[0038] Silent Window This is the inherent silence window between PTT release and the next PTT press, during which no audio service is transmitted but the radio link remains active. At that time, the terminal is A new round of key negotiation is completed: the local end regenerates a temporary private key via CSPRNG, calculates a new temporary public key, and sends it to the peer end via the activated wireless link. The peer end also regenerates a temporary private key and a temporary public key, completes the ECDH shared secret calculation, and uses the latest key from both sides. and current KDF is re-executed for the parameters to derive a new session key; the new session key is cached in local secure storage, and the current session key remains valid until the current call ends; the cached new session key is used directly when the next call is established, skipping the complete negotiation process during the establishment phase. During this period, pre-negotiation is not performed within the silent window, the current session key continues to be used, and the wireless link bandwidth of the silent window is fully reserved.

[0039] Duration of silent window If the pre-negotiation process is insufficient to complete the full ECDH negotiation (e.g., the operator presses keys rapidly and continuously), the pre-negotiation process will not be initiated, and the system will wait for the next silent window that meets the required duration. If, during the waiting process, the CFO dual-rate judgment mechanism triggers key revocation first and the buffer is empty, the terminal will perform emergency key negotiation using a compressed handshake method (reusing the exchanged temporary public key to complete the shared secret calculation, omitting the public key regeneration step) to ensure that the latency required for communication recovery meets the real-time requirements of the walkie-talkie service.

[0040] In a group call scenario, the group coordinator (usually the calling terminal that initiated the call) first collects the key data reported by each of the n member terminals in the group. Before reporting, each member terminal encrypts its local key gene using the group coordinator's temporary public key to ensure secure transmission. The group coordinator concatenates all member terminal key genes and injects them into the KDF as an additional entropy source to generate the group session key. Then, using a threshold-based secret sharing scheme to... Divide into n shared shares Set the reconstruction threshold to a, satisfying Each share is distributed after being encrypted with the temporary public key of the corresponding member terminal. The shares held by any 'a' terminals can be recovered through Lagrange interpolation. If the number of shares is less than a, then no information can be obtained regarding... Any information. Group key refresh uses a triggering strategy based on the lowest KS member: any member terminal in the group calculates the KS locally. At that time, a group key refresh request is initiated to the group coordinator, and this request is made through the current... The encrypted key is broadcast; upon receiving the refresh request, the group coordinator initiates a new round of group key synthesis and distribution, using the current call round's silent window as the execution window, to re-collect the latest key genes from each member's terminal and generate a new key. The keys are then redistributed using a threshold scheme. The key strength assessment value (KS) obtained by the member terminal with the lowest KS value is used as the criterion for group key refresh. This means that the overall security policy of the group is determined by the terminal with the lowest security margin, preventing a single weak terminal from lowering the overall security level of the group undetected.

[0041] When the call ends, the terminal performs a one-way hash chain rolling update on the key gene, using the current key gene as input, and calculates the next round of key gene through a one-way hash function: ; Where k is the call round index. The key gene used for the k-th call. This is the sequence number for this call. It is a one-way hash function. The introduction of this feature means that the input to each round of the hash chain not only depends on the key gene of the previous round, but is also bound to the sequence number of the current call. The evolution trajectory of the hash chain between different call rounds is thus affected. They are strictly incremental and independent of each other, and cannot be intercepted in a particular round. Backtracking This ensures the forward security of historical sessions. After the hash chain is updated, the terminal clears the current session key from memory using a secure erase method (resetting it to zero after multiple random overwrites). ,make sure It does not remain in memory or accessible storage areas. The key gene for the next call is persistently stored in the terminal's secure storage area, so that it can participate in the construction of a new key gene when the next call is established.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A walkie-talkie communication method based on dynamic keys, characterized in that, include: When a call is established, PTT behavior parameters of both communicating parties are collected and quantified to obtain the behavior entropy component. A statistical stability assessment is performed on the historical carrier frequency offset sequences of each device to obtain the frequency offset baseline. Using behavioral entropy components and frequency offset baselines as key genes, a session key bound to the call behavior and hardware characteristics of both parties is generated through a key negotiation protocol and key derivation. During the call, a comprehensive assessment of the security margin of the session key is conducted, and the negotiation and caching of the new session key are completed in advance within the PTT silent window; Simultaneously, the real-time changes in carrier frequency offset are continuously monitored and compared with the frequency offset baseline according to the rate of change. If the rate of change is within the normal range, the session key remains valid. When an abnormal mutation is detected, the current session key is revoked. If a new session key is already stored in the cache, it is immediately enabled to restore communication; otherwise, key negotiation is re-initiated. When the call ends, the key gene is updated using a one-way hash chain, and the current session key is cleared. The comprehensive evaluation uses the key strength evaluation value KS to quantify the security margin of the session key, calculated at each PTT release time: ； Where q represents the cumulative number of encrypted packets in this session, and SNR represents the measured signal-to-noise ratio of the current channel. As the initial reference signal-to-noise ratio, The channel power-law stress coefficient; Let i be the entropy component of the current behavior, and i be the terminal index. The mean entropy of historical behavior. This represents the deviation of the current behavior entropy component from the historical mean. This is the stress sensitivity coefficient for abnormal behavior; This represents the cumulative baseline drift. This is the cumulative stress coefficient for carrier frequency offset drift. To preset a safety lower limit threshold; when KS is lower than At that time, a new session key pre-negotiation is triggered within the silent window; At the end of each call, calculate the current... and The deviation amount, and with Historical variance estimates Comparison: When , Normal update when the preset tolerance multiple is used. ; in The forgetting factor has a value of ,make Recent behavior is given a higher weight, while long-term behavior is given an exponentially decaying weight. when At that time, freeze The update will not change the current The call was included in the mean calculation and marked as an abnormal behavior event, triggering a reassessment of KS.

2. The method according to claim 1, characterized in that, The PTT behavior parameters include three components: button duration, Delay from the moment the key is pressed to the moment the voice activity detection is triggered. The short-time energy mean of the first frame of audio signal after voice activity detection is triggered. ; Key press duration Delay with short-time energy mean The three components constitute the PTT behavior vector for this call, which is accumulated by sliding within the acquisition window of K consecutive calls to form a behavior matrix. , where K is the preset acquisition window length.

3. The method according to claim 2, characterized in that, The behavioral entropy component is processed using the permutation entropy method on the behavioral matrix. Quantization, behavioral entropy components Defined as: ； Where i is the terminal index. The permutation entropy value is calculated for the row matrix with embedding dimension m. This represents the theoretical maximum value of the permutation entropy under embedding dimension m. The normalized value range is within (0,1) for the behavior entropy component; the embedding dimension m is a preset parameter.

4. The method according to claim 3, characterized in that, The key gene is constructed through a cascaded hash operation, and the key gene of terminal i... Defined as: ； in, It is a one-way hash function. Indicates byte concatenation; This is the device's inherent identifier entropy value. This is the timestamp for this call. The channel noise sample value, This is the call sequence number. This is the output after the behavioral entropy components have been independently hashed. It is a scalar normalized to the carrier frequency offset measurement value after rounding with a fixed quantization step size.

5. The method according to claim 4, characterized in that, The key negotiation protocol adopts the elliptic curve Diffie-Hellman protocol, and the key derivation adopts a hash-based key derivation function; The calling terminal i and the called terminal j each generate a temporary private key using a cryptographically secure random number generator. , And calculate the corresponding temporary public key. , where G is the base point of the elliptic curve; After exchanging temporary public keys, both parties calculate the shared key separately. ; Using the key genes of both parties Call timestamps serve as a source of additional entropy. Salt value, call sequence number For contextual information, derive the session key using the key derivation function: ； in It is injected as an additional entropy source and does not participate in the generation of temporary private keys.

6. The method according to claim 1, characterized in that, The rate of change comparison employs a dual-rate judgment mechanism: The preamble synchronization frame carrying the measured carrier frequency offset each time the PTT button is pressed carries the actual value. , where t is the measurement order index; The frequency offset baseline is denoted as The receiver calculates the measured value and the frequency offset baseline. Deviation amount: ； And calculate the short-time mutation rate between two adjacent measurements: ； in This represents the deviation obtained from the previous measurement; when Exceeding the short-term mutation rate threshold If the event is detected as an abnormal mutation, the session key will be revoked. when Exceeding the long-term cumulative deviation threshold but Not exceeding At that time, the frequency offset baseline was adjusted using the average of nearly W measurements. Perform adaptive updates and record the cumulative baseline drift. ; when The session key remains valid during this period.

7. The method according to claim 1, characterized in that, The silence window is the inherent silence time window between the release of PTT and the next PTT press. There is no audio service transmission within this window, but the wireless link remains active; when At that time, the terminal is A new round of key negotiation is completed within the current call, and the new session key is cached locally. The current session key remains valid until the end of the current call, and the cached new session key is used directly when the next call is established. During this period, pre-negotiation is not performed within the silent window, and the current session key continues to be used.

8. The method according to claim 1, characterized in that, In a group call scenario, the group coordinator collects the key data reported by each of the n member terminals in the group, and uses a threshold secret sharing scheme to split the group session key into n shared shares. A reconstruction threshold of 'a' is set, and each share is encrypted and distributed to the corresponding terminal. Shares from any 'a' terminals can be used to reconstruct the group session key. Any member terminal within the group can determine locally that KS is lower than 'a'. When a request for a group key refresh is sent to the group coordinator, the group coordinator initiates a new round of group key synthesis and distribution within the silent window of the current call round; the KS obtained by the member terminal with the lowest key strength evaluation value is used as the basis for determining the group key refresh.

9. The method according to claim 1, characterized in that, At the end of the call, a one-way hash chain is rolled over to update the key gene. The terminal uses the current key gene as input and calculates the next key gene using a one-way hash function. ； Where i is the terminal index. The key gene used for the k-th call. This is the call sequence number. It is a one-way hash function; after the update is completed, the session key is immediately cleared in memory.

Images