A side information aware finite field multiplexing retransmission and evaluation method for multiple unicast scenarios

Abstract

The application relates to a side information perception finite field multiplex retransmission and evaluation method for multiple unicast scenes, belonging to the technical field of wireless communication, and aims to solve the problems of different user equipment demand set heterogeneity and possible increase of unknown load detection of non-demand data packets in a reliable retransmission scene where multiple unicast sessions coexist. The application constructs a demand matrix and a three-value state feedback matrix, which are used for distinguishing target missing data packets, buffered side information data packets and non-demand unknown data packets; then, candidate retransmission data packets are selected according to target request weights, and the translatability of a concurrent retransmission data packet set across unicast sessions is checked in combination with the side information state of each target user equipment and unknown detection load; after the translatability constraint is met, the base station maps the selected data packets to finite field elements and performs multiplex transmission; and the user equipment uses the locally buffered data packets as side information to complete detection and on-demand data recovery in a limited candidate space.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology and relates to a method for data retransmission and evaluation in multi-user reliable transmission. Background Technology

[0002] With the development of Vehicle-to-Everything (V2X), Internet of Things (IoT), Industrial Internet, and mobile multimedia services, the number of user devices simultaneously accessing wireless systems continues to increase. In many practical applications, different user devices do not always request the same set of data packets, but rather belong to different service sessions with significantly different required data content. For example, multiple vehicles may request different road condition updates, and multiple sensors or terminals may request different control information or application data. This type of service can be abstracted as a reliable transmission scenario where multiple unicast sessions coexist. Due to factors such as wireless channel fading, noise, obstruction, and interference, the reception status of each user device's target data packet differs after the initial transmission, thus requiring efficient retransmission.

[0003] In multiple unicast scenarios, retransmission design needs to consider not only whether a user equipment (UE) has received a data packet, but also whether that data packet belongs to the UE's demand set. Unlike broadcast or multicast scenarios where all UEs are directed to the same set of source data packets, user demands in multiple unicast scenarios are heterogeneous: a data packet may be the target request for some UEs, but a non-demand data packet for other UEs. Under the on-demand reception model, UEs typically only cache data packets that they have requested and successfully received, without additionally caching non-demand data packets, to reduce terminal power consumption, storage overhead, and feedback maintenance complexity. Therefore, although non-demand data packets do not contribute to the UE's retransmission benefits, they may increase its unknown detection load when transmitted concurrently across sessions.

[0004] Traditional Automatic Repeat Request (ARQ) mechanisms typically retransmit individual data packets. While simple to implement, they struggle to leverage the differences in side information between different user devices. In multiple unicast sessions, ARQ needs to serve a large number of target request pairs between user devices and data packets one by one, resulting in significant completion latency and average decoding latency.

[0005] Random Linear Network Coding (RLNC) can improve throughput through linear combinations of finite fields, but the receiver typically needs to accumulate a sufficient number of linearly independent coded packets before joint decoding. This mechanism has non-instantaneous decoding characteristics and requires operations such as matrix inversion or Gaussian elimination, making it difficult to meet the needs of user equipment in latency-sensitive services to quickly recover its target data packets.

[0006] Instantly Decodable Network Coding (IDNC) utilizes cached data packets at the receiving end as side information, constructing decodeable encoded packets via binary XOR. However, traditional IDNC typically requires the target user equipment to have at most one data packet missing from the encoded packet. In multiple unicast sessions, due to the heterogeneity of demand sets, the non-demand data packets not being stored as long-term side information, and the inconsistency between target requests and the detection of unknown payloads, the encodeable combination space of traditional IDNC is further limited.

[0007] While finite-domain multiplexing offers higher information carrying capacity for single-round retransmissions, the key challenge in multiple unicast sessions lies not merely in finite-domain mapping, but in determining which data packets can be sent simultaneously across sessions. Specifically, a data packet might be a target request for some user devices, but a non-demand unknown packet for others; the former brings retransmission benefits, while the latter does not, but increases the detection candidate space for the target user devices. If treated as a broadcast or multicast model, all unbuffered data packets would be considered as missing targets, making it impossible to distinguish between target requests and non-demand unknown loads, and hindering the establishment of feedback update rules suitable for on-demand unicast services. Therefore, reliable retransmission of multiple unicast packets requires a redesign of the scheduling mechanism around the demand matrix, the three-valued state feedback matrix, and cross-session translatability checks. Summary of the Invention

[0008] This invention aims to solve the problems of heterogeneous user equipment demand sets and the potential increase in unknown detection load for non-demand data packets in reliable retransmission scenarios where multiple unicast sessions coexist, as well as the difficulty of traditional retransmission methods in fully utilizing cross-session concurrency opportunities.

[0009] A side-information-aware finite-domain multiplexing and retransmission method for multiple unicast scenarios includes:

[0010] S1. The base station constructs a demand matrix based on the service requirements of each user equipment, and constructs a three-valued state feedback matrix based on the reception status feedback after the initial transmission; the three-valued state feedback matrix is ​​used to distinguish between three states: target data packet has been cached, target data packet is still missing, and non-demand data packet.

[0011] S2. The base station determines, based on the three-valued state feedback matrix, that there are still active data packets that have not met the target request; for the test concurrent data packet set composed of active data packets, it determines the target user equipment corresponding to the test concurrent data packet set, and determines whether the test concurrent data packet set satisfies the cross-session translatability constraint based on the actual number of unknown data packets in the test concurrent data packet set for each target user equipment;

[0012] S3. The base station determines the target request weight based on the number of unmet target requests corresponding to each active data packet, selects candidate data packets according to the target request weight, and adds them to the test concurrent data packet set in sequence; when the test concurrent data packet set simultaneously satisfies the maximum concurrent data packet number constraint on the base station side and the cross-session translatability constraint, the added candidate data packets are retained; otherwise, the candidate data packets are skipped, thereby obtaining the current round of concurrent retransmission data packet set.

[0013] S4. The base station allocates finite field element pairs to the data packets in the concurrent retransmission data packet set, maps the data packet bits to finite field elements, superimposes finite fields and patterns on the finite fields to generate finite fields and patterns, and maps the finite fields and patterns to complex plane transmission symbols for transmission; the target user equipment uses its own cached target data packets as side information to construct a restricted candidate finite field and pattern set, performs detection within the restricted candidate finite field and pattern set, and recovers the target data packets belonging to its own demand set;

[0014] S5. Each target user equipment feeds back the target data packet recovery result to the base station. The base station updates the three-valued state feedback matrix according to the feedback result and performs the next round of retransmission until the target requests of all user equipment are satisfied. Among them, the target user equipment only adds the target data packets that belong to its own demand set and have been successfully recovered to the cache set.

[0015] Furthermore, the three-valued state feedback matrix Its elements are defined as:

[0016] ,

[0017] Where t represents the first... Rotational passing, Indicates user equipment The target data packet has been successfully received and cached. , For packets in the set of source packets involved in a unicast session; Indicates user equipment The target data packet is still missing and needs to be recovered. ; Indicates data packet Not a user equipment The set of demands; Indicates user equipment Required data package set and for In the The set of cached target data packets and the set of target data packets to be recovered after the round. This represents the difference operation.

[0018] Furthermore, the process described in S2 for determining whether the set of concurrent test data packets satisfies the cross-session translatability constraint includes:

[0019] For user equipment The set of currently concurrently retransmitted data packets that belong to the local unrecovered set of its target request. , This is a set of test concurrent data packets consisting of active data packets; if Then the user equipment It is the first The target receiver for retransmission in this round is not the target receiver for this round; according to... Determine the target user equipment set for the current round ;

[0020] The cross-session translatability constraint is satisfied as follows: User equipment Detecting unknown payload in the current set of concurrent retransmitted packets. ; The translatability threshold.

[0021] Furthermore, in determining whether the set of concurrent test packets satisfies the cross-session translatability constraint, the current round's concurrent subset should also satisfy the base station-side concurrent quantity constraint: , The maximum number of concurrent data packets is preset on the base station side, where k represents the k-th source data packet. Indicates the total number of source data packets; For the first Round-robin scheduling indicator variable, hour, =1, otherwise 0.

[0022] Furthermore, the process by which the base station in S3 determines the target request weight based on the number of unmet target requests corresponding to each active data packet includes:

[0023] In the At the start of round-robin retransmission, the base station reads the three-valued state feedback matrix obtained from the previous round of updates. And count the number of unfulfilled target requests corresponding to each data packet in each column: For data packets The base station initializes its target request weight to 0, and then sequentially checks the three-valued state feedback matrix. Each element in the column: when When, it indicates that the data packet Belongs to user equipment The demand set that has not yet been obtained by the user equipment will send the data packet. The target request weight increases by 1; when or At that time, the target request weight is not increased; thus, the data packet In the The target request weight at the start of the round is ;like If the data packet is not currently an active target request, it will not participate in this round of candidate scheduling; If the packet is selected, it will be placed in the candidate queue.

[0024] Furthermore, when the test concurrent data packet set simultaneously satisfies both the maximum concurrent data packet count constraint on the base station side and the cross-session translatability constraint, the process of retaining the added candidate data packets includes:

[0025] During the candidate check phase, the base station will select candidate data packets. Temporarily add to the current concurrent subset to obtain the test set. The target user equipment set corresponding to the test set is For any user equipment Its detection of unknown load in the test set is , Indicates the first -1 round of retransmission unknown set Send a set to the base station;

[0026] If the candidate data packets satisfy the cross-session translatability constraint, The base station then accepts the candidate data packet and... Otherwise, skip the candidate data packet; this process continues until the current concurrent subset reaches the maximum number of concurrent data packets, or the candidate queue has been traversed.

[0027] Furthermore, in S4, during the process of the base station allocating finite field element pairs to data packets in the concurrent retransmission data packet set, a translatability determination is required, including:

[0028] During the construction of the concurrent retransmission data packet set in step S3, each time a candidate data packet is added to the test concurrent data packet set, a translatability determination is performed on the test concurrent data packet set: the number of data packets in the test concurrent data packet set does not exceed the maximum number of concurrent data packets on the base station side, and for each target user equipment corresponding to the test concurrent data packet set, its detected unknown load does not exceed a preset translatability threshold; thus, the concurrent retransmission data packet set output after concurrency in step S3 is... satisfy as well as .

[0029] Furthermore, the process by which the base station allocates finite field element pairs to packets in the set of concurrent retransmission packets in S4 includes:

[0030] Let the set of concurrent retransmission data packets output after step S3 be... The base station can select a finite field based on the concurrency order. And each concurrent data packet will be in the first... Bits at each bit position Mapped to finite field elements, forming multiplexed emission symbols ; in receiving symbols After that, user equipment The first candidate can be determined by performing maximum likelihood detection within the candidate space. Bits at each bit position.

[0031] Furthermore, in S5, each target user equipment feeds back the target data packet recovery result to the base station, and the base station updates the three-valued state feedback matrix based on the feedback result and performs the next round of retransmission, including:

[0032] No. After the round-robin retransmission is completed, each user equipment verifies its own target recovery result and sends back ACK / NACK; if the target user equipment... Successfully recovered If the target data packet is in the middle, then update:

[0033] ,

[0034] Determine user equipment The actual unknown set of data packets in the current concurrent retransmission data packet set. ,for Unknown data packets not required by the user equipment are not added to the physical layer detection process, even if the user equipment obtains a bit estimate of the packet. And it still maintains in the three-valued state feedback matrix Status; if the user equipment is not the target receiver in this round, or has not successfully recovered the target data packet, its buffer set and the set to be recovered remain unchanged;

[0035] The base station updates the three-valued state feedback matrix based on feedback from all user equipment and the demand matrix. The process then proceeds to the next round of encoding and retransmission; the retransmission process ends when all user equipment's target requests have been satisfied, and its termination condition is:

[0036] ,

[0037] in, This indicates the completion delay of multiple unicast session retransmission processes and counts the number of encoded retransmission rounds required after the initial transmission ends.

[0038] A side-information-aware finite-domain multiplexing and retransmission evaluation method for multiple unicast scenarios, wherein the evaluation is based on average decoding latency, including:

[0039] Calculate the average recovery rounds for target requests that remain unmet after all initial transmissions have ended; for any initial target request... If the unicast session involves data packets in the source data packet set , for The initial set of target data packets to be recovered is denoted as... For user equipment Data packets successfully recovered Number of encoding retransmission rounds, average decoding latency:

[0040] ,

[0041] in, This indicates the number of user devices.

[0042] Beneficial effects:

[0043] This invention enables cross-session concurrent retransmission scheduling across multiple unicast sessions, allowing the selection of retransmitted data packets to simultaneously consider the target request benefit and the target user equipment's detection of unknown load. Compared to packet-by-packet retransmission, this invention improves the ability of a single-round retransmission to serve the target request; compared to traditional binary real-time decodable network coding, this invention expands the feasible concurrent combination space through finite-domain multiplexing mapping and side-information-assisted detection, thereby reducing completion latency and average decoding latency. Attached Figure Description

[0044] Figure 1 This is a comparison curve showing the change in the completion delay of multiple unicast systems as the probability of demand varies in an embodiment of the present invention.

[0045] Figure 2This is a comparison curve showing the average decoding latency of multiple unicast systems as a function of demand probability in an embodiment of the present invention.

[0046] Figure 3 This is a comparison curve showing the change in the number of source data packets as the completion delay of multiple unicast systems under different retransmission signal-to-noise ratios in this embodiment of the invention.

[0047] Figure 4 This is a comparison curve showing the average decoding delay of multiple unicast systems varying with the number of source data packets under different retransmission signal-to-noise ratios in this embodiment of the invention.

[0048] Figure 5 This is a comparison curve showing the change in the number of source data packets as the completion delay of multiple unicast systems under different initial packet loss rates in this embodiment of the invention.

[0049] Figure 6 This is a comparison curve showing the average decoding latency of multiple unicast systems varying with the number of source data packets under different initial packet loss rates in embodiments of the present invention. Detailed Implementation

[0050] This invention addresses reliable retransmission scenarios with multiple coexisting unicast sessions. It tackles the challenges of heterogeneous user equipment demand sets, the potential for increased detection load from non-demand data packets, and the difficulty of traditional retransmission methods in fully utilizing cross-session concurrency opportunities. The invention proposes a side-information-aware finite-domain multiplexing retransmission and evaluation method for multiple unicast scenarios, comprising the following steps:

[0051] S1. The base station constructs a demand matrix based on the service requirements of each user equipment, and constructs a three-valued state feedback matrix based on the reception status feedback after the initial transmission; the three-valued state feedback matrix is ​​used to distinguish between three states: target data packet is cached, target data packet is still missing, and non-demand data packet.

[0052] S2. The base station determines, based on the three-valued state feedback matrix, that there are still active data packets that have not met the target request; for the test concurrent data packet set composed of the active data packets, it determines the target user equipment corresponding to the test concurrent data packet set, and determines whether the test concurrent data packet set satisfies the cross-session translatability constraint based on the actual number of unknown data packets in the test concurrent data packet set for each target user equipment.

[0053] S3. The base station determines the target request weight based on the number of unmet target requests corresponding to each active data packet, selects candidate data packets according to the target request weight, and adds them to the test concurrent data packet set in sequence. When the test concurrent data packet set simultaneously satisfies the maximum number of concurrent data packets on the base station side and the cross-session translatability constraint, the added candidate data packets are retained; otherwise, the candidate data packets are skipped, thereby obtaining the current round of concurrent retransmission data packet set.

[0054] S4. The base station allocates finite field element pairs to the data packets in the concurrent retransmission data packet set, maps the data packet bits to finite field elements, superimposes finite fields and patterns on the finite fields to generate finite fields and patterns, and maps the finite fields and patterns to complex plane transmission symbols for transmission; the target user equipment uses its own cached target data packets as side information to construct a restricted candidate finite field and pattern set, performs detection within the restricted candidate finite field and pattern set, and recovers the target data packets belonging to its own demand set.

[0055] S5. Each target user equipment feeds back the target data packet recovery result to the base station. The base station updates the three-valued state feedback matrix according to the feedback result and performs the next round of retransmission until the target requests of all user equipment are satisfied. Among them, the target user equipment only adds the target data packets that belong to its own demand set and have been successfully recovered to the cache set, and does not add non-demand data packets that do not belong to its own demand set to the cache set.

[0056] Furthermore, the method also includes a theoretical evaluation step S6:

[0057] S6. Under ideal error-free conditions, based on the active data packets that still have unmet target requests after the initial transmission, the initial target request weight of each active data packet, and the total number of target requests, a feasible schedule is constructed that simultaneously satisfies the data packet coverage constraint, the maximum concurrent data packet number constraint on the base station side, and the cross-session translatability constraint. The ideal recovery round for each active data packet is determined based on the feasible schedule, and the completion delay and average decoding delay are calculated based on the ideal recovery round for each active data packet, serving as a theoretical reference boundary for evaluating the delay performance of the retransmission method.

[0058] Where there is no conflict, the embodiments and technical features described in this specification can be combined with each other. For those skilled in the art, equivalent substitutions or modifications made to the finite field order, element pair selection method, constellation mapping method, candidate data packet sorting criteria, feedback period, or caching strategy without departing from the inventive concept should all fall within the protection scope of this invention. The following description, in conjunction with specific embodiments, will further illustrate these embodiments. Specific implementation method one:

[0060] The side-information perception finite-domain multiplexing, retransmission, and evaluation method for multiple unicast scenarios described in this embodiment mainly includes:

[0061] S1, Multiple unicast system models and ternary state feedback:

[0062] This implementation corresponds to the reliable retransmission process after a base station sends data packets belonging to different unicast sessions to multiple user equipments. Let the base station be BS, and the set of user equipment be... ,in This indicates the number of user devices; the unicast system is abbreviated as "system", and the set of source data packets involved in all unicast sessions within the system is denoted as "system". ,in This represents the total number of source packets, and the length of each source packet is [length value missing]. Bit.

[0063] In multiple unicast scenarios, different user devices may request different data packets. For user devices... The set of required data packets is denoted as To uniformly describe the demand relationship between user equipment and data packets, a demand matrix is ​​defined. :

[0064]

[0065] Therefore, user equipment The set of requirements can be represented as:

[0066]

[0067] In one implementation, the system employs an on-demand caching model. The user equipment (UE) only caches data packets belonging to its own demand set that have been successfully received or recovered. In subsequent encoding and retransmission rounds, the cached data packets are used as side information to offset the contribution of known data packets to the finite field and to constrain candidate finite fields and pattern sets. For non-demand data packets that do not belong to its demand set, even if their decision is obtained during the current round of physical layer joint detection, the UE does not cache these non-demand data packets and does not use them as side information in subsequent encoding and retransmission rounds. Therefore, In the The set of cached target data packets after the round and the set of target data packets to be recovered satisfy:

[0068]

[0069] in, This represents the difference operation.

[0070] During the initial transmission phase, the base station sends a set The source data packets are received and verified by the user equipment, which then reports the reception status back to the base station via the feedback link. Based on this, the base station constructs a three-valued state feedback matrix. Its elements are defined as:

[0071]

[0072] in, Indicates user equipment The target data packet has been successfully received and cached. ; Indicates user equipment The target data packet is still missing and needs to be recovered. ; Indicates data packet Not a user equipment The set of demands. When all user equipment requests the complete set of data packets, the three-valued state feedback matrix no longer appears. The element, the system model can degenerate into a broadcast or multicast retransmission model.

[0073] To describe data packets that a user equipment cannot directly cancel during finite-domain multiplexing detection, an unknown set is defined. :

[0074]

[0075] The unknown set includes target data packets that the user equipment still needs to recover and non-demand data packets that do not belong to its demand set. Non-demand data packets are not included in the completion delay and average decoding delay, but if they are concurrently scheduled with target data packets in the same round, they will increase the candidate detection space of the target user equipment.

[0076] S2, Cross-session concurrent retransmission and translatability conditions:

[0077] In the At the start of round-coded retransmission, the base station uses the three-valued state feedback matrix obtained from the previous round of feedback. Select the set of packets for concurrent retransmission in the current round. First, only packets that still have unmet target requests need to participate in scheduling. Define the active packet set:

[0078]

[0079] Base station from Select several data packets to form the current round's concurrent retransmission data packet set:

[0080]

[0081] in, This represents the concurrency order of the current round. Set the maximum number of concurrent data packets for the base station side.

[0082] Unlike a single broadcast session, the current round of concurrent retransmissions of data packets is a set of... Different data packets can serve different target user equipment in different unicast sessions. Therefore, the base station selects several data packets from the aforementioned set of active data packets to form the current round of concurrent retransmission data packet set. The process is called cross-session concurrent scheduling.

[0083] For user equipment The local set to be recovered that belongs to its target request in the current set of concurrent retransmission data packets is defined as:

[0084]

[0085] Meanwhile, user equipment The set of packets that are actually unknown in the current set of concurrently retransmitted packets is defined as:

[0086]

[0087] like Then the user equipment It is the first The target receiver for round-robin retransmission; if Then the user equipment Not being the target receiver in this round, it will not participate in the decoding revenue statistics for this round. In many unicast on-demand reception models, there are typically... That is, the target data packet to be recovered is only a part of the actual unknown detection data packet.

[0088] according to Define the target user equipment set for the current round:

[0089]

[0090] To control the detection complexity and reliability of target user equipment, the number of actually unknown data packets in the current concurrent subset for any target user equipment shall not exceed a preset translatable threshold. The corresponding cross-session translatability constraints are:

[0091]

[0092] The above formula shows that, in multiple unicast scenarios, the constraint is not on all user devices, nor on the target set to be restored. The constraint is not on the size of the data packet itself, but on the total number of data packets that are actually unknown to the target user equipment in the current concurrent subset. This constraint allows the base station to improve cross-session concurrency capabilities while avoiding the target user equipment facing an excessively large candidate detection space.

[0093] To facilitate matrix implementation, the first... Round-robin scheduling indicator variables:

[0094]

[0095] User equipment In the Whether a wheel is the target receiver can be represented as:

[0096]

[0097] in, This is an indicator function; it takes the value 1 if the condition is met, and 0 if the condition is not met.

[0098] As can be seen from the definition of detecting unknown loads,

[0099]

[0100] in, Indicates user equipment Detecting unknown payloads within the current set of concurrent retransmitted packets, compared to the actual set of unknown packets. The bases are equal. At the same time, Used to indicate user equipment Does it belong to the current round of target user equipment set: when hour, ;when hour, .

[0101] Therefore, the aforementioned cross-session translatability constraints using set cardinality representation Introducing target user equipment indicator variables and detecting unknown loads Then, it can be equivalently represented as

[0102]

[0103] Among them, when At that time, user equipment For the target user equipment in the current round, its detection of unknown load should meet the following requirements: ;when At that time, user equipment It is not the target user equipment in the current round and does not participate in the recovery of the target data packets in the current round; therefore, there is no need to impose the aforementioned constraint on its detection of unknown load. The above condition is equivalent to

[0104]

[0105] Furthermore, the current round of concurrent subsets should also satisfy the base station-side concurrent quantity constraint:

[0106] .

[0107] S3, cross-session greedy concurrent scheduling:

[0108] In the At the start of round-robin retransmission, the base station reads the three-valued state feedback matrix obtained from the previous round of updates. And then, column by column, count the number of unfulfilled target requests corresponding to each data packet. Specifically, for each data packet... The base station initializes its target request weight to 0, and then sequentially checks the three-valued state feedback matrix. Each element in the column: when When, it indicates that the data packet Belongs to user equipment The demand set that has not yet been obtained by the user equipment will send the data packet. The target request weight increases by 1; when or At that time, the target request weight is not increased. Therefore, the data packet... In the The target request weight at the start of the round is

[0109]

[0110] in, This indicates that the request is still ongoing and the data packet has not yet been received. The number of user devices. If If the data packet is not currently an active target request, it will not participate in this round of candidate scheduling; If the packet is selected, it will be placed in the candidate queue.

[0111] During the candidate check phase, the base station will select candidate data packets. Temporarily add it to the current concurrent subset to obtain the test set:

[0112]

[0113] The target user device set corresponding to the test set is:

[0114]

[0115] For any The unknown load detected in the test set is:

[0116]

[0117] The conditions under which a candidate data packet can be accepted are:

[0118]

[0119] If the above formula holds true and The base station then accepts the candidate data packet and... Otherwise, skip the candidate data packet. This process continues until the current concurrent subset reaches the maximum number of concurrent data packets, or the candidate queue has been traversed.

[0120] The core of the greedy scheduling algorithm described above is that the scheduling benefit of a data packet depends solely on the state where the target is missing. The decision is made based on the fact that the feasibility check considers both the target missing data packets and non-required unknown data packets. Therefore, the method avoids sending a large number of non-required unknown data packets concurrently to the same target user device, which could overload the detection process.

[0121] S4. Cross-session multiplexing of transmission and on-demand recovery:

[0122] After the cross-session greedy concurrent scheduling in step S3, the base station obtains the set of concurrent retransmission data packets for the current round. In step S3, during the construction of the concurrent retransmission data packet set, a translatability determination is performed on the test concurrent data packet set each time a candidate data packet is added. The translatability determination specifically includes: the number of data packets in the test concurrent data packet set does not exceed the maximum number of concurrent data packets on the base station side; and for each target user equipment corresponding to the test concurrent data packet set, its detected unknown load does not exceed a preset translatability threshold. Therefore, the concurrent retransmission data packet set output in step S3... satisfy as well as The base station outputs the set of concurrent retransmission data packets in step S3. Limited-domain multiplexing transmission is performed. Unlike the transmission of data packets to the same target set in a broadcast scenario, the concurrent data packets in this embodiment can come from multiple unicast sessions, and each target user equipment only recovers the target data packets belonging to its own demand set.

[0123] The "translatability determination" actually includes two explicit conditions:

[0124] Maximum concurrent data packet count constraint on the base station side;

[0125] The current round detects unknown load constraints for each target user device.

[0126] set up The base station can select a finite field based on the concurrency order. And each concurrent data packet will be in the first... Bits at each bit position Mapped to elements of a finite field, forming a multiplexed emission symbol:

[0127]

[0128] in, Indicates the data packet to be assigned finite field element mapping, This represents the mapping of a finite field and its elements to constellation points in the complex plane.

[0129] The above formula is only used to illustrate the multiplexing transmission process on a cross-session concurrent set; the finite field order, element pairs, and constellation mapping can be configured according to the modulation capabilities supported by the system.

[0130] For target user equipment The cached target data packets provide long-term side information, while non-demand data packets do not provide long-term side information. Therefore, this user equipment in the first... Wheels targeting Constructing candidate bit space from unknown data packets:

[0131]

[0132] In receiving symbols Afterwards, the target user equipment Perform maximum likelihood detection within the candidate bit vector space:

[0133]

[0134] in, Indicates the target user equipment In the The first round of re-passes At each bit position, the estimated bit vector of the actual unknown data packet obtained by maximum likelihood detection has each component corresponding to a set of actual unknown data packets. Each data packet in the first... Bit estimation at each bit position, i.e. , Indicates a set of actual unknown data packets Each data packet in the first... The candidate bit vector space is formed by the combination of candidate bits at each bit position; Indicates in the target user device Known side information set After being fixed, the candidate bit vector A defined finite field and a pattern; This represents a mapping function that maps finite fields and patterns to complex plane constellation points.

[0135] After the target user equipment completes the detection, it only uses the estimated vector Extract the target set to be recovered. The corresponding bit estimate is used to recover the target data packets belonging to its own requirement set; the difference between the actual unknown data packet set and the target set to be recovered. Unknown data packets that are not in demand will not be added to the on-demand cache set, even if they are jointly decided during the detection process.

[0136] S5. Feedback Update and Termination Conditions:

[0137] No. After the round-robin retransmission is completed, each user equipment verifies its own target recovery result and sends back ACK / NACK. If the target user equipment... Successfully recovered If the target data packet is in the middle, then update:

[0138]

[0139] for Unknown data packets not required by the user equipment are not added to the physical layer detection process, even if the user equipment obtains a bit estimate of the packet. And it still maintains in the three-valued state feedback matrix Status. If the user equipment is not the target receiver in this round, or has failed to recover the target data packet, its buffer set and the set to be recovered remain unchanged.

[0140] The base station base station bases the base station base stations based on feedback from all user equipment and the demand matrix. Update the ternary state feedback matrix Then, the next round of encoding and retransmission begins. The retransmission process ends when all user equipment's target requests have been satisfied, with the termination condition being:

[0141]

[0142] in, This indicates the completion delay of multiple unicast session retransmission processes and counts the number of encoded retransmission rounds required after the initial transmission ends.

[0143] For the average decoding latency, calculate the average recovery rounds for target requests that have not been satisfied after all initial transmissions have ended. For any initial target request... ,like ,remember For user equipment Data packets successfully recovered If the number of encoding retransmission rounds is given, then the average decoding latency is defined as:

[0144]

[0145] If all target requests have been satisfied after the initial transmission, i.e. Then, both the completion latency and the average decoding latency are defined as 0.

[0146] Average decoding latency As a performance evaluation metric for retransmission methods, it is used to measure the average number of rounds in which each initially unmet target request is recovered during the retransmission process.

[0147] S6. Theoretical Evaluation Reference Boundary:

[0148] To evaluate the latency performance of the cross-session scheduling method described in this invention, in one embodiment, a theoretical reference evaluation model under ideal error-free conditions is further established. It should be noted that the subscripts... Indicates an inter-session context; subscript or superscript. It is an abbreviation for Reference, indicating that the corresponding variable belongs to the ideal error-free theoretical reference model; superscript This represents the optimal value achieved among all theoretically feasible schedules. Step S5 does not include... The marked delay variable represents the delay index in the actual retransmission process.

[0149] First, define the set of data packet indices that still need to participate in scheduling during the initial encoding and retransmission phase as follows:

[0150]

[0151] in, This indicates that at least one set of data packet indices remains unsatisfied after the initial transmission has ended.

[0152] For any packet index Its initial target request weight is defined as

[0153] in, This indicates that the request for data packets has ended but the packets have not yet been received. The number of user devices.

[0154] The total number of all initially unmet target requests is

[0155] like This indicates that all user equipment's target requests have been satisfied after the initial transmission, and at this point, both the actual latency metric and the theoretical reference latency metric are defined as 0. The following section addresses... The situation will be explained.

[0156] Under ideal error-free conditions, theoretical reference scheduling variables are used. Indicates data packet Was it arranged in the Each encoded retransmission slot is sent, i.e.

[0157]

[0158] Under ideal error-free conditions, as long as the set of concurrent data packets in a given retransmission slot meets the cross-session translatability condition, all target user equipment corresponding to that slot can successfully recover their target data packets. Therefore, each belonging to The data packets only need to be sent once, and the corresponding data packet coverage constraint is...

[0159] The number of data packets concurrently scheduled in each retransmission time slot does not exceed the maximum number of concurrent data packets on the base station side; therefore, the concurrency constraint on the base station side is:

[0160] For the Each retransmission time slot, the corresponding set of target user equipment is defined as follows:

[0161]

[0162] For any target user device In its first The detected unknown load in each retransmission time slot must not exceed a preset translatable threshold; therefore, the translatability constraint for the target user equipment is:

[0163]

[0164] In the above translatable constraints, the state The target data packets that the corresponding user equipment still needs to recover, status Both correspond to unknown data packets that are not required; both will increase the unknown detection load on the target user device.

[0165] Let the set of all theoretical reference scheduling variables that satisfy the binary variable constraint, packet coverage constraint, base station-side concurrency constraint, and target user equipment translatability constraint be denoted as .

[0166] Among them, subscript This indicates the inter-session scenario corresponding to the feasible scheduling set, with the subscript... This indicates that the feasible scheduling set belongs to the ideal error-free theoretical reference model.

[0167] For a given theoretically feasible schedule data packets The theoretical reference recovery round is determined by the retransmission time slot it is scheduled to receive, i.e.

[0168] Given the theoretically feasible schedule, its theoretical reference completion delay is expressed as:

[0169] Given the theoretically feasible schedule, the theoretical reference average decoding latency for all initially unmet target requests is expressed as:

[0170]

[0171] Based on the initial target request weight By definition, the above theoretical reference average decoding latency can also be equivalently expressed as:

[0172]

[0173] Under ideal error-free conditions, for any condition satisfying The initial target request, when the data packet When a data packet is scheduled to be sent and the current set of concurrent data packets satisfies the cross-session translatability constraint, the corresponding user equipment can recover the data packet in that time slot. Therefore, requesting the same data packet Each target user equipment has the same theoretical reference recovery round, which is... Therefore, the initial target request weight can be used. A weighted average is calculated based on the theoretical reference recovery rounds for each active data packet.

[0174] make Then retransmission packet by packet can guarantee that no more than The scheduling of all active data packets is completed within each retransmission time slot. Therefore, the optimal theoretical completion delay reference bound in the inter-session scenario is defined as...

[0175]

[0176] The optimal theoretical average decoding latency reference bound in inter-session scenarios is defined as follows:

[0177]

[0178] It should be noted that the optimal theoretical completion delay reference bound and the optimal theoretical average decoding delay reference bound correspond to two different optimization objectives. The former is used to find the minimum number of retransmission slots required to complete all target requests; the latter is used to minimize the average recovery rounds for all initially unmet target requests. Therefore, the optimal scheduling order corresponding to the two does not have to be the same.

[0179] Actual average decoding delay in step S5 Based on the actual recovery rounds requested by each target during the actual retransmission process Calculation; in this step Based on the theoretical reference recovery rounds for each data packet under ideal error-free conditions Both methods evaluate the same target request that remains unfulfilled after the initial transmission ends, but they use different sources for recovery rounds. These are performance metrics generated during the actual retransmission process. It is a theoretical reference bound obtained under ideal error-free and globally optimal scheduling conditions. It is only used to evaluate the latency performance of actual cross-session retransmission methods and does not participate in the actual retransmission scheduling, finite field multiplexing transmission, target data packet recovery or feedback update process in steps S1 to S5.

[0180] Example: Simulation verification of multiple unicast retransmission performance:

[0181] To verify the technical effectiveness of the method described in this invention in multiple unicast scenarios, this embodiment verifies the latency performance under unified simulation conditions with different demand probabilities, different source data packet sizes, and different initial packet loss rates. The simulation is only used to illustrate the feasibility and effect trends of this invention and does not constitute a limitation on the scope of protection of this invention.

[0182] First, we examine the impact of demand probability on system latency. Demand probability is defined as follows: , indicating user equipment Request data packet The probability of . When When the size is small, the set of target requests from different user devices is relatively sparse; as the size increases... As the number of requests increases, the density of target requests in the system gradually increases. At that time, all user devices request all data packets, and multiple unicast models degenerate into broadcast or multicast scenarios.

[0183] In demand probability simulation, the number of user equipment is taken as... The number of source data packets is taken Demand probability The initial packet loss rate changes from 0.2 to 1.0. Signal-to-noise ratio during retransmission stage The data packet length is taken as bits, maximum concurrent data packets . Figure 1 and Figure 2 Results on the completion delay and average decoding delay of multiple unicast systems as a function of demand probability are presented.

[0184] Depend on Figure 1 and Figure 2 It is evident that the latency of the packet-by-packet retransmission scheme increases significantly with the increase in demand probability. This is because packet-by-packet retransmission can only retransmit a single data packet per round, and its performance is mainly affected by the initial total number of target requests. In contrast, the present invention can maintain low latency under different demand probabilities. This is because the present invention can utilize finite-domain multiplexing to carry multiple data packets during cross-session scheduling and limit the detection of unknown loads through target user equipment translatability checks, thereby expanding the feasible combinations of concurrent retransmissions.

[0185] Further examine the number of source data packets Impact on retransmission performance. In this set of simulations, the number of user equipment is taken as... The probability of demand is taken Number of source data packets from Change to 55, data packet length is taken as... bits, maximum concurrent data packets . Figure 3 and Figure 4 The system completion delay and average decoding delay are given under the conditions of fixed initial packet loss rate and demand probability, and different retransmission signal-to-noise ratios. The result of the change.

[0186] Depend on Figure 3 and Figure 4 It can be seen that as the number of source data packets increases, the completion delay and average decoding delay of each scheme generally increase. This is because, with a fixed number of user devices, demand probability, and initial packet loss rate, the number of initial target requests in the system increases approximately linearly with the number of source data packets. This invention achieves low latency under different retransmission signal-to-noise ratios, and when the signal-to-noise ratio increases during the retransmission stage, the reliability of finite-domain multiplexing detection is enhanced, further reducing the completion delay and average decoding delay.

[0187] Figure 5 and Figure 6 The following figures illustrate different initial packet loss rates under the conditions of fixed retransmission stage signal-to-noise ratio and demand probability. The impact of completion latency and average decoding latency on performance as the number of source data packets increases. Figure 5 and Figure 6 It can be seen that, in the same When the initial packet loss rate is high, the overall completion latency and average decoding latency of each scheme increase. This is because a higher initial packet loss rate generates more unmet target requests, increasing the number of user equipment and data packet request pairs that need to be recovered during the encoding and retransmission phase. Under both initial packet loss rate conditions, the present invention performs worse than packet-by-packet retransmission and traditional real-time decodable network coding schemes, indicating that it can effectively utilize concurrent retransmission opportunities between multiple unicast sessions.

[0188] comprehensive Figures 1 to 6 As can be seen, this invention characterizes the target request and non-demand unknown states in multiple unicast scenarios through a demand matrix and a ternary state feedback matrix. Combined with side-information-assisted detection and cross-session greedy scheduling, it improves the information carrying capacity of single-round retransmissions while ensuring the translatability constraint of the target user equipment. Therefore, this invention can reduce completion latency and average decoding latency in multiple unicast scenarios with heterogeneous demand sets, varying initial packet loss rates, and expanding source data packet sizes.

[0189] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A finite-domain multiplexing and retransmission method for side-information perception in multiple unicast scenarios, characterized in that, include: S1. The base station constructs a demand matrix based on the service requirements of each user equipment, and constructs a three-valued state feedback matrix based on the reception state feedback after the initial transmission. The three-valued state feedback matrix is ​​used to distinguish between three states: target data packet has been cached, target data packet is still missing, and non-required data packet. S2. The base station determines, based on the three-valued state feedback matrix, that there are still active data packets that have not met the target request; for the test concurrent data packet set composed of active data packets, it determines the target user equipment corresponding to the test concurrent data packet set, and determines whether the test concurrent data packet set satisfies the cross-session translatability constraint based on the actual number of unknown data packets in the test concurrent data packet set for each target user equipment; S3. The base station determines the target request weight based on the number of unmet target requests corresponding to each active data packet, selects candidate data packets according to the target request weight, and adds them to the test concurrent data packet set in sequence. When the test concurrent data packet set simultaneously satisfies the maximum concurrent data packet number constraint on the base station side and the cross-session translatability constraint, the added candidate data packets are retained; otherwise, the candidate data packets are skipped, thereby obtaining the current round of concurrent retransmission data packet set. S4. The base station allocates finite field element pairs to the data packets in the concurrent retransmission data packet set, maps the data packet bits to finite field elements, superimposes finite fields and patterns on the finite fields to generate finite fields and patterns, and maps the finite fields and patterns to complex plane transmission symbols for transmission; the target user equipment uses its own cached target data packets as side information to construct a restricted candidate finite field and pattern set, performs detection within the restricted candidate finite field and pattern set, and recovers the target data packets belonging to its own demand set; S5. Each target user equipment feeds back the target data packet recovery result to the base station. The base station updates the three-valued state feedback matrix according to the feedback result and performs the next round of retransmission until the target requests of all user equipment are satisfied. Among them, the target user equipment only adds the target data packets that belong to its own demand set and have been successfully recovered to the cache set.

2. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 1, characterized in that, The three-valued state feedback matrix Its elements are defined as: ， Where t represents the first... Rotational passing, Indicates user equipment The target data packet has been successfully received and cached. , For packets in the set of source packets involved in a unicast session; Indicates user equipment The target data packet is still missing and needs to be recovered. ; Indicates data packet Not a user equipment The set of demands; Indicates user equipment Required data package set and for In the The set of cached target data packets and the set of target data packets to be recovered after the round. This represents the difference operation.

3. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 2, characterized in that, The process described in S2 for determining whether the set of concurrent test data packets satisfies the cross-session translatability constraint includes: For user equipment The set of currently concurrently retransmitted data packets that belong to the local unrecovered set of its target request. , This is a set of test concurrent data packets consisting of active data packets; if Then the user equipment It is the first The target receiver for retransmission in this round is not the target receiver for this round; according to... Determine the target user equipment set for the current round ; The cross-session translatability constraint is satisfied as follows: User equipment Detecting unknown payload in the current set of concurrent retransmitted packets. ; The translatability threshold.

4. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 3, characterized in that, In determining whether the test concurrent data packet set satisfies the cross-session translatability constraint, the current round's concurrent subset should also satisfy the base station-side concurrent quantity constraint: , The maximum number of concurrent data packets is preset on the base station side, where k represents the k-th source data packet. Indicates the total number of source data packets; For the first Round-robin scheduling indicator variable, hour, =1, otherwise 0.

5. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 3, characterized in that, The process by which the base station determines the target request weight based on the number of unmet target requests corresponding to each active data packet, as described in S3, includes: In the At the start of round-robin retransmission, the base station reads the three-valued state feedback matrix obtained from the previous round of updates. And count the number of unfulfilled target requests corresponding to each data packet in each column: For data packets The base station initializes its target request weight to 0, and then sequentially checks the three-valued state feedback matrix. Each element in the column: when When, it indicates that the data packet Belongs to user equipment The demand set that has not yet been obtained by the user equipment will send the data packet. The target request weight increases by 1; when or At that time, the target request weight is not increased; thus, the data packet In the The target request weight at the start of the round is ;like If the data packet is not currently an active target request, it will not participate in this round of candidate scheduling; If the packet is selected, it will be placed in the candidate queue.

6. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 5, characterized in that, S3, when the test concurrent data packet set simultaneously satisfies the maximum concurrent data packet count constraint on the base station side and the cross-session translatability constraint, the process of retaining the added candidate data packets includes: During the candidate check phase, the base station will select candidate data packets. Temporarily add to the current concurrent subset to obtain the test set. The target user equipment set corresponding to the test set is For any user equipment Its detection of unknown load in the test set is , Indicates the first -1 round of retransmission unknown set Send a set to the base station; If the candidate data packets satisfy the cross-session translatability constraint, The base station then accepts the candidate data packet and... Otherwise, skip the candidate data packet; this process continues until the current concurrent subset reaches the maximum number of concurrent data packets, or the candidate queue has been traversed.

7. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 6, characterized in that, In S4, the base station needs to perform a translatability determination during the process of allocating finite field element pairs to data packets in the concurrent retransmission data packet set, including: During the construction of the concurrent retransmission data packet set in step S3, each time a candidate data packet is added to the test concurrent data packet set, a translatability determination is performed on the test concurrent data packet set: the number of data packets in the test concurrent data packet set does not exceed the maximum number of concurrent data packets on the base station side, and for each target user equipment corresponding to the test concurrent data packet set, its detected unknown load does not exceed a preset translatability threshold; thus, the concurrent retransmission data packet set output after concurrency in step S3 is... satisfy as well as .

8. The side information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios according to claim 7, characterized in that, In S4, the process by which the base station allocates finite field element pairs to packets in the set of concurrent retransmission packets includes: Let the set of concurrent retransmission data packets output after step S3 be... The base station can select a finite field based on the concurrency order. And each concurrent data packet will be in the first... Bits at each bit position Mapped to finite field elements, forming multiplexed emission symbols ; in receiving symbols After that, user equipment The first candidate can be determined by performing maximum likelihood detection within the candidate space. Bits at each bit position.

9. A finite-domain multiplexing and retransmission method for side-information perception in multiple unicast scenarios according to claim 7, characterized in that, In S5, each target user equipment feeds back the target data packet recovery result to the base station. The process by which the base station updates the three-valued state feedback matrix based on the feedback result and performs the next round of retransmission includes: No. After the round-robin retransmission is completed, each user equipment verifies its own target recovery result and sends back ACK / NACK; if the target user equipment... Successfully recovered If the target data packet is in the middle, then update: ， Determine user equipment The actual unknown set of data packets in the current concurrent retransmission data packet set. ,for Unknown data packets not required by the user equipment are not added to the physical layer detection process, even if the user equipment obtains a bit estimate of the packet. And it still maintains in the three-valued state feedback matrix Status; if the user equipment is not the target receiver in this round, or has not successfully recovered the target data packet, its buffer set and the set to be recovered remain unchanged; The base station updates the three-valued state feedback matrix based on feedback from all user equipment and the demand matrix. The process then proceeds to the next round of encoding and retransmission; the retransmission process ends when all user equipment's target requests have been satisfied, and its termination condition is: ， in, This indicates the completion delay of multiple unicast session retransmission processes and counts the number of encoded retransmission rounds required after the initial transmission ends.

10. A method for evaluating side-information perception finite-domain multiplexing and retransmission in multiple unicast scenarios, characterized in that, The side-information sensing finite-domain multiplexing and retransmission method for multiple unicast scenarios as described in any one of claims 1 to 9 is evaluated using average decoding latency, including: Calculate the average recovery rounds for target requests that remain unmet after all initial transmissions have ended; for any initial target request... If the unicast session involves data packets in the source data packet set , for The initial set of target data packets to be recovered is denoted as... For user equipment Data packets successfully recovered Number of encoding retransmission rounds, average decoding latency: ， in, This indicates the number of user devices.

Images