Hybrid fair scheduling method and system based on wireless optical communication network

Abstract

The invention relates to the technical field of wireless optical communication, and discloses a hybrid fair scheduling method and system based on a wireless optical communication network. Time slots are dynamically distributed by comprehensively considering user channel capacity and throughput requirements on the basis of different throughput and time delay requirements of interactive users and transmission users; meanwhile, a scheduling structure is reconstructed according to different service types; for the interactive users, through an iteration mechanism of preferentially allocating a time slot to the interactive user with the maximum service guiding priority, and through presetting the minimum reallocation time slot, resources are rapidly obtained, it is ensured that the time slot to be recombined is preferentially consumed, and time delay accumulation caused by waiting for the long time slot of the transmission type user is avoided; and for the transmission type users, after the time slots of the interactive type users are distributed, the time slots are distributed to the transmission type users in a centralized manner according to the service oriented priority, so that the throughput loss caused by context switching is reduced, the requirement of completing large-volume data transmission within several seconds is met, and the network real-time performance and the resource utilization rate are integrally improved.

Description

Technical Field

[0001] This invention relates to the field of wireless optical communication technology, and in particular to a hybrid fair scheduling method and system based on wireless optical communication networks. Background Technology

[0002] As people's demand for network data increases, obtaining information online faster and conveniently without a fixed location has become an unavoidable optimization issue. Among them, optical wireless communication (OWC) modulates parameters such as the intensity, color, or polarization state of wireless optical signals to transmit signals to users faster. It has advantages such as large transmission bandwidth, strong anti-interference ability, no need for electromagnetic frequency licenses, and green environmental protection.

[0003] The applications of wireless optical communication are extensive. For example, in large supermarkets, wireless light sources above people can provide communication for users holding devices while simultaneously offering illumination, alleviating the problem of unstable and slow network connections in such large supermarkets. In underground parking lots, the introduction of wireless optical communication modules can improve visibility and address poor signal strength in underground communication service areas. In mechanized smart factories, wireless optical communication enables intelligent equipment operation, rapid command issuance, and intelligent area division, thereby improving operational efficiency. In railway trains or underground passenger transport, wireless optical communication can enhance network connectivity, preventing users from being unable to access the internet when trains travel at high speeds. These are all pain points that human activities need to address; therefore, research on wireless optical communication technology is crucial.

[0004] Time slot scheduling in wireless optical communication has significant practical implications. The principle of scheduling technology is to allocate resource blocks such as frequency, time, and power to individual users. However, different allocation methods yield different results. In a typical indoor OWC scenario, multiple users within the same room form a network, where most users should share similar quality of service. However, as users move further away from the wireless light source, channel quality degrades, leading to significant differences in achievable data rates. Therefore, traditional scheduling schemes may result in unsatisfactory resource allocation and service performance. One of the main optimization objectives is to allocate different time and frequency resource blocks to different users through effective algorithms to achieve goals in terms of user demand, fairness among users, or user throughput. Since Time-Division Multiplexing (TDM) technology is widely used in OWC networks due to its simplicity and economy, the allocation of time resource blocks is even simpler and more convenient.

[0005] Wireless optical communication based on a wireless light source and photodetector (PD) framework uses a wireless light source as the signal transmitter and a PD as the signal receiver. The wireless light source transmits optical signals through free space, and the PD receives the transmitted optical signals. The PD converts the received optical signals into electrical signals, then demodulates the electrical signals to recover the transmitted data. Existing wireless optical communication scheduling algorithms and their shortcomings are as follows:

[0006] The Proportional Fairness (PF) scheduling algorithm divides time resources into many equal time slots. During scheduling, it's necessary to determine the ownership of each time slot. When determining the ownership of a particular time slot, the ratio of the instantaneous transmission rate of all users to their cumulative throughput is used as the scheduling priority for each user. The user with the highest priority is assigned that time slot and transmission occurs within it. As a user acquires more consecutive time slots, their cumulative throughput gradually increases, causing their priority to decrease. Conversely, for other users who haven't acquired time slots, their cumulative throughput gradually decreases, leading to an increase in priority and a better chance of acquiring time slots. Over time, a balance is achieved between the throughput of all users and the fairness among them. Traditional proportional fairness algorithms, when user channel conditions vary significantly, result in users with poor channel conditions being in a prolonged state of starvation compared to those with better channel conditions. PF guarantees that the throughput obtained by each user is proportional to their average channel condition; that is, users with good channel conditions should receive higher throughput, and users with good channel conditions are prioritized in scheduling. However, this technique cannot consider the actual needs of users.

[0007] Round-Robin (RR) scheduling algorithms define a polling period during which all users are scheduled sequentially, with the total number of scheduling slots equal to the polling period size. Without weighting, all users have the same priority, and each user receives an equal time slot for transmission. With weighting, the transmission time slots are allocated based on user weights; different weighting methods lead to different allocation results. The round-robin scheduling algorithm ignores the actual needs and current state of all users, simply allocating equal resources (such as time blocks) to each user in a loop. This is unfair in terms of perceived performance. Furthermore, the same user needs to wait for a polling period between each scheduling cycle; even if data packets arrive, they cannot be sent immediately, resulting in a poor experience for users with high latency requirements.

[0008] Scheduling algorithms based on the Maximum Carrier-to-Interference Ratio (MAX C / I) prioritize scheduling for users with the best signal quality, aiming for maximum throughput. The C / I ratio is the power ratio of the useful signal to the interfering signal. A higher C / I ratio indicates better signal quality and a lower Bit Error Rate (BER) for a given coding and modulation scheme. Higher signal quality allows for more efficient modulation and higher coding rates, resulting in higher data transmission rates on the same time-frequency resources. MAX C / I scheduling prioritizes users with the best signal quality, aiming for maximum throughput. However, this algorithm can cause significant unfairness. Users located at the edge of a wireless optical cell, due to greater distance, path loss, and stronger interference, consistently have a lower C / I value. Under this algorithm, they almost never receive scheduling resources, potentially rendering their services (such as calls and internet access) completely unavailable. Their effective service area is limited to the central area of ​​the cell with good channel conditions, resulting in a small scheduling coverage.

[0009] Therefore, PF, RR, and MAX C / I primarily aim to optimize throughput and fairness, or to strike a trade-off between them. Among these, the PF algorithm achieves a balance between throughput and fairness, providing a practical compromise compared to the RR and MAX C / I algorithms, which focus on maximizing fairness and throughput respectively. Despite these advancements, existing scheduling strategies often neglect the service needs of individual users. Maxim-minimum fairness scheduling prioritizes the needs of the least privileged users, striving to maximize the allocation of the smallest user's share. Once the user with the smallest need is satisfied, if the remaining users cannot be satisfied, they receive the average of the remaining resources. Maxim-minimum fairness scheduling is highly inefficient; resources may be over-allocated to users with poor channel conditions, leading to low transmission efficiency. Furthermore, it fails to differentiate user priorities, failing to provide more resources for high-level users or critical services with higher transmission demands. While max-minimum fairness can meet users' minimum data rate requirements, resource utilization efficiency decreases if user needs are diverse.

[0010] Besides throughput and fairness, latency is also a critical service metric that requires optimization. Many existing methods attempt to mitigate excessive latency caused by scheduling imbalances by using compensation factors. For example, some researchers have introduced a method to assign compensation factors to users experiencing high latency to increase their scheduling priority; for instance, compensation factor-based scheduling algorithms add compensation factors to the priority calculation of traditional proportional fairness algorithms to compensate for specific situations, such as compensating the user with the highest latency to obtain higher priority and thus more scheduling opportunities, thereby reducing latency; or compensating the user with the lowest rate to allow them to receive more scheduling, thereby achieving greater throughput. While adding compensation factors to scheduling methods changes the priority calculation value based on traditional proportional fairness, it primarily alters the order of priority, and the numerical improvement may not be significant. Furthermore, it only considers the relative latency ranking between users without explicitly relating it to absolute latency values, resulting in suboptimal latency performance.

[0011] In summary, existing scheduling methods in wireless optical communication networks generally suffer from problems such as neglecting individual user service needs, incomplete optimization of key indicators such as fairness, throughput, and latency, and low resource utilization efficiency. These methods are difficult to adapt to complex and diverse real-world application scenarios, affect the experience of latency-sensitive services, and lead to a decrease in resource utilization in wireless optical communication networks. Summary of the Invention

[0012] Therefore, the technical problem to be solved by the present invention is to overcome the problems of low resource utilization and poor fairness among users in wireless optical communication networks caused by neglecting the individual service needs of users in the prior art.

[0013] To address the aforementioned technical problems, this invention provides a hybrid fair scheduling method based on a wireless optical communication network, comprising: S1: Based on the latency sensitivity and throughput requirements of each user's data to be transmitted, all users are divided into transmission users and interaction users; S2: Real-time acquisition of the real-time throughput demand and real-time channel capacity of all users in the wireless optical communication network within the current scheduling period; S3: For each user, calculate the ratio of real-time demand throughput to real-time channel capacity, obtain the real-time demand time slot, and normalize it to obtain the normalized demand factor; based on the product of the normalized demand factor and the preset cycle period, obtain the maximum allocable time slot; select the smaller value between the maximum allocable time slot and the real-time demand time slot as the actual allocated time slot for each user. S4: Obtain the ratio of the actual allocated time slots for each user to the preset minimum time slot for reallocation, and use it as the number of time slots to be reassembled for each user; S5: Calculate the average real-time throughput demand of each user in the historical cycle period as the historical average throughput, and combine it with the real-time channel capacity to obtain the initial priority of each user. S6: Based on the real-time channel capacity of each user and the initial demand throughput of the current scheduling period, construct the initial demand time slots of each user, and calculate the average to obtain the global average demand factor of the current scheduling period. S7: Perform a logarithmic transformation on the ratio of each user's real-time demand slot to the global average demand factor of the current scheduling period, multiply it by the initial priority of each user, and obtain the service orientation priority of each user. S8: Obtain the interactive user with the highest service orientation priority, allocate a time slot to the interactive user, and decrease the number of time slots to be reassembled for the interactive user by one; S9: Return to step S2 until all interactive users have 0 slots to be reassembled. Then, allocate slots to transport users in descending order of service-oriented priority.

[0014] Preferably, for each user, the ratio of real-time demand throughput to real-time channel capacity is calculated to obtain the real-time demand time slot, and then normalized to obtain the normalized demand factor, including: Calculate the real-time throughput requirements of each user With real-time channel capacity The ratio is used to obtain the real-time demand slots for each user. , is represented as: ; Normalize the real-time demand slots of each user to obtain the normalized demand factor for each user. , is represented as: ; in, Indicates the first The user in the first Demand throughput within a scheduling cycle Indicates the first The user in the first Channel capacity within a scheduling period, Indicates the first The user in the first Normalized demand factor within each scheduling cycle; , This indicates the total number of users.

[0015] Preferably, the smaller value between the maximum allocatable time slot for each user and the real-time demand time slot is selected as the actual allocated time slot for each user, expressed as: ; in, Indicates the first The user in the first The actual allocated time slots within each scheduling cycle; Indicates the first The user in the first The maximum allocatable time slot within a scheduling period is expressed as follows: , This indicates the preset cycle period.

[0016] Preferably, the average real-time throughput demand of each user in the historical cycle is calculated as the historical average throughput. Combined with the real-time channel capacity, the initial priority of each user is obtained, expressed as: ; in, Indicates the first The user in the first Initial priority within a scheduling cycle, Indicates the first The user in the first Real-time channel capacity within a scheduling cycle Indicates the first The user in the first The average real-time throughput demand in the historical cycle of each scheduling cycle; This indicates the preset weighting factor.

[0017] Preferably, based on the real-time channel capacity of each user and the initial demand throughput of the current scheduling period, the initial demand time slots of each user are constructed, and the average is calculated to obtain the global average demand factor of the current scheduling period, including: Based on the real-time channel capacity of each user Initial demand throughput for the current scheduling period The initial demand slots for each user are constructed, represented as follows: ; Calculate the average of the initial demand slots for all users, and use this as the global average demand factor for the current scheduling period, expressed as: ; Among them, the The user in the first Initial demand throughput within a scheduling cycle For the first The user in the first Real-time throughput required when the first time slot is allocated within a scheduling cycle.

[0018] Preferably, the ratio of each user's real-time demand slot to the global average demand factor of the current scheduling period is logarithmically transformed, and multiplied by the initial priority of each user to obtain the service orientation priority of each user, expressed as: ; in, Indicates the first The user in the first Service-oriented priority within a scheduling cycle.

[0019] Preferably, time slots are allocated to transport users in descending order of service orientation priority, including: calculating the number of time slots to be reassembled for each transport user, and allocating the corresponding number of time slots to be reassembled for each transport user in descending order of service orientation priority.

[0020] Preferably, the number of time slots to be reassembled for each transmission type user is calculated, including: Obtain the remaining unallocated time slots for each transmission user after all interactive users have 0 time slots to be reassembled in the current scheduling period; Calculate the ratio of the remaining unallocated time slots for each transmission type user to the preset minimum time slot for reallocation, and use this ratio as the number of time slots to be reassembled for each transmission type user.

[0021] Preferably, after all transport-type users have been allocated time slots, the process includes: In the current scheduling cycle, an initial transmission queue is constructed based on the time order in which all interactive users are allocated the first time slot; All transport users are added to the initial transport queue in descending order of service orientation priority, and then the actual transport queue is obtained. Obtain the actual allocated time slots for each interactive user and transmission user, and form a time slot matrix.

[0022] This invention provides a hybrid fair scheduling system based on a wireless optical communication network, comprising: Integrated into the transmitter: The source module acquires the data to be transmitted from each user; The central control module, based on the hybrid fair scheduling method based on the wireless optical communication network as described above, obtains the actual transmission queues and time slot matrices for all users' transmissions. The light source driving module generates driving signals based on the actual transmission queue and time slot matrix of all users. Wireless light source, which emits corresponding light signals based on driving signals; Each user at the receiving end demodulates and extracts the optical signal to obtain the transmission result of the corresponding data to be transmitted.

[0023] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0024] The hybrid fair scheduling method based on wireless optical communication networks described in this invention calculates the real-time demand time slots for each user based on their real-time demand throughput and real-time channel capacity within the wireless optical communication network during the current scheduling period. This calculation is then normalized to obtain a normalized demand factor, transforming the demand intensity of different users into a normalized demand factor of the same dimension. This ensures that users with high demand receive more resources, while users with low demand do not occupy unnecessary resources. Simultaneously, the maximum allocable time slot is obtained based on the normalized factor and a preset cycle period. The smaller value between the maximum allocable time slot and the real-time demand time slot is selected as the actual allocated time slot, ensuring that no user experiences insufficient throughput due to channel fluctuations, nor wastes time slots due to low demand.

[0025] This invention calculates the service orientation priority of each user by performing a logarithmic transformation on the ratio of each user's real-time demand slot to the global average demand factor of the current scheduling period. This enhances the priority of users with high demand while avoiding excessive resource skew towards any one user. As user throughput demands are gradually met, their priority naturally decreases, thus prioritizing scheduling for users with higher but unmet demands. This ensures fairness and comparability among users, better meeting the service needs of multiple users with similar service quality levels while maintaining fairness and reducing average latency.

[0026] This invention addresses the different throughput and latency requirements of interactive and transmission users by dynamically allocating time slots based on a comprehensive consideration of user channel capacity and throughput requirements. Simultaneously, it restructures the scheduling structure according to different service types. For interactive users, an iterative mechanism prioritizes allocating time slots to users with the highest service orientation priority. By pre-setting a minimum reallocation time slot, resources are quickly acquired, ensuring that their waiting time slots are consumed first, avoiding latency accumulation caused by waiting for long time slots from transmission users. For transmission users, after time slot allocation for interactive users is completed, time slots are centrally allocated to transmission users according to service orientation priority, reducing throughput loss due to context switching and adapting to their need to complete large-volume data transmission within seconds. This invention, through small-granularity time slot reorganization based on a pre-set minimum reallocation time slot and a scheduling logic prioritizing interactive users, specifically reduces transmission latency for interactive users while simultaneously reducing context switching latency for transmission users, thereby improving overall network real-time performance and resource utilization. Attached Figure Description

[0027] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart of the steps of the hybrid fair scheduling method based on wireless optical communication network of the present invention; Figure 2This is a schematic diagram of an indoor multi-user downlink wireless optical communication network structure; Figure 3 This is a block diagram of an indoor multi-user downlink wireless optical communication network system; Figure 4 This is a comparison chart of the joint fairness index of various scheduling algorithms under different user demand loads; Figure 5 This is a comparison chart of the joint fairness index of various scheduling algorithms under different user channel capacities; Figure 6 This is a comparison chart of user throughput for various scheduling algorithms under different loads; Figure 7 This is a comparison chart of waiting latency for different users under various scheduling algorithms when the demand load is 0.5. Figure 8 This is a comparison chart of the average latency of various scheduling algorithms under different demand loads. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0029] Reference Figure 1 The flowchart shown illustrates the steps of the hybrid fair scheduling method based on a wireless optical communication network of the present invention. The specific steps include: S1: Based on the latency sensitivity and throughput requirements of each user's data to be transmitted, all users are divided into transmission users and interaction users; S2: Real-time acquisition of the real-time throughput demand and real-time channel capacity of all users in the wireless optical communication network within the current scheduling period; S3: For each user, calculate the ratio of real-time demand throughput to real-time channel capacity, obtain the real-time demand time slot, and normalize it to obtain the normalized demand factor; based on the product of the normalized demand factor and the preset cycle period, obtain the maximum allocable time slot; select the smaller value between the maximum allocable time slot and the real-time demand time slot as the actual allocated time slot for each user. S4: Obtain the ratio of the actual allocated time slots for each user to the preset minimum time slot for reallocation, and use it as the number of time slots to be reassembled for each user; S5: Calculate the average real-time throughput demand of each user in the historical cycle period as the historical average throughput. Combined with the real-time channel capacity, obtain the initial priority of each user, expressed as: ; in, Indicates the first The user in the first Initial priority within a scheduling cycle, Indicates the first The user in the first Real-time channel capacity within a scheduling cycle Indicates the first The user in the first The average real-time throughput demand in the historical cycle of each scheduling cycle; Indicates the preset weighting factor; S6: Based on the real-time channel capacity of each user and the initial demand throughput of the current scheduling period, construct the initial demand time slots for each user, and calculate the average to obtain the global average demand factor for the current scheduling period, including: Based on the real-time channel capacity of each user Initial demand throughput for the current scheduling period The initial demand slots for each user are constructed, represented as follows: ; Calculate the average of the initial demand slots for all users, and use this as the global average demand factor for the current scheduling period, expressed as: ; Among them, the The user in the first Initial demand throughput within a scheduling cycle For the first The user in the first Real-time throughput required when the first time slot is allocated within a scheduling cycle. S7: The ratio of each user's real-time demand slot to the global average demand factor of the current scheduling period is logarithmically transformed, and multiplied by the initial priority of each user to obtain the service orientation priority of each user, expressed as: ; S8: Obtain the interactive user with the highest service orientation priority, allocate a time slot to the interactive user, and decrease the number of time slots to be reassembled for the interactive user by one; S9: Return to step S2 until all interactive users have 0 slots to be reassembled. Then, allocate slots to transport users in descending order of service-oriented priority.

[0030] This invention calculates the service orientation priority of each user by performing a logarithmic transformation on the ratio of each user's real-time demand slot to the global average demand factor of the current scheduling period. This enhances the priority of users with high demand while avoiding excessive resource skew towards any one user. As user throughput demands are gradually met, their priority naturally decreases, thus prioritizing scheduling for users with higher but unmet demands. This ensures fairness and comparability among users, better meeting the service needs of multiple users with similar service quality levels while maintaining fairness and reducing average latency.

[0031] Specifically, in step S3, the actual allocated time slots for each user are obtained, including: S3-1: Calculate the real-time throughput requirements of each user. With real-time channel capacity The ratio is used to obtain the real-time demand slots for each user. , is represented as: ; S3-2: Normalize the real-time demand slots for each user to obtain the normalized demand factor for each user. , is represented as: ; S3-3: Based on the normalized demand factors and preset cycle periods of each user The product of these factors yields the maximum allocatable time slot for each user. The expression is ; S3-4: Select the smaller value between the maximum allocatable time slot for each user and the real-time demand time slot as the actual allocated time slot for each user. ; in, Indicates the first The user in the first Demand throughput within a scheduling cycle Indicates the first The user in the first Channel capacity within a scheduling period, Indicates the first The user in the first Normalized demand factor within each scheduling cycle; , This indicates the total number of users.

[0032] The hybrid fair scheduling method based on wireless optical communication networks described in this invention calculates the real-time demand time slots for each user based on their real-time demand throughput and real-time channel capacity within the wireless optical communication network during the current scheduling period. This calculation is then normalized to obtain a normalized demand factor, transforming the demand intensity of different users into a normalized demand factor of the same dimension. This ensures that users with high demand receive more resources, while users with low demand do not occupy unnecessary resources. Simultaneously, the maximum allocable time slot is obtained based on the normalized factor and a preset cycle period. The smaller value between the maximum allocable time slot and the real-time demand time slot is selected as the actual allocated time slot, ensuring that no user experiences insufficient throughput due to channel fluctuations, nor wastes time slots due to low demand.

[0033] Specifically, in step S8, time slots are allocated to transport users in descending order of service orientation priority. This includes: calculating the number of time slots to be reassembled for each transport user, and allocating the corresponding number of time slots to be reassembled for each transport user in descending order of service orientation priority. Calculating the number of time slots to be reassembled for each transport user includes: obtaining the remaining time slots to be allocated for each transport user after all interactive users have zero time slots to be reassembled in the current scheduling period; and calculating the ratio of the remaining time slots to be allocated for each transport user to the preset minimum time slot for reallocation, which is used as the number of time slots to be reassembled for each transport user.

[0034] In this embodiment, after all transport users have been allocated time slots, the process includes: constructing an initial transport queue based on the time order in which all interactive users were allocated their first time slot during the current scheduling period; adding all transport users to the initial transport queue in descending order of service orientation priority to obtain the actual transport queue; and obtaining the actual time slots allocated to each interactive user and transport user to form a time slot matrix. Based on the actual transport queue and time slot matrix constructed in this embodiment, this invention provides a hybrid fair scheduling system based on a wireless optical communication network, including: Integrated into the transmitter: The source module acquires the data to be transmitted from each user; The central control module, based on the hybrid fair scheduling method based on the wireless optical communication network as described above, obtains the actual transmission queues and time slot matrices for all users' transmissions. The light source driving module generates driving signals based on the actual transmission queue and time slot matrix of all users. Wireless light source, which emits corresponding light signals based on driving signals; Each user at the receiving end demodulates and extracts the optical signal to obtain the transmission result of the corresponding data to be transmitted.

[0035] This invention addresses the different throughput and latency requirements of interactive and transmission users by dynamically allocating time slots based on a comprehensive consideration of user channel capacity and throughput requirements. Simultaneously, it restructures the scheduling structure according to different service types. For interactive users, an iterative mechanism prioritizes allocating time slots to users with the highest service orientation priority. By pre-setting a minimum reallocation time slot, resources are quickly acquired, ensuring that their waiting time slots are consumed first, avoiding latency accumulation caused by waiting for long time slots from transmission users. For transmission users, after time slot allocation for interactive users is completed, time slots are centrally allocated to transmission users according to service orientation priority, reducing throughput loss due to context switching and adapting to their need to complete large-volume data transmission within seconds. This invention, through small-granularity time slot reorganization based on a pre-set minimum reallocation time slot and a scheduling logic prioritizing interactive users, specifically reduces transmission latency for interactive users while simultaneously reducing context switching latency for transmission users, thereby improving overall network real-time performance and resource utilization.

[0036] Based on the above embodiments, in this embodiment, as... Figure 2 In the indoor multi-user downlink wireless optical communication network shown, time slot allocation is performed using the hybrid fair scheduling method provided by this invention; for example... Figure 2 As shown, this network system is equipped with wireless light sources as access points (APs), and under the coordination of a central controller, it provides services to multiple user devices (User Equipment). i (i=1,2,3…) provides downlink communication services. Each user equipment (AP) is equipped with a photodetector (PD) to receive optical signals and feeds back Channel State Information (CSI) and throughput requirements to the central controller in real time via an uplink mechanism. The channel gain between the AP and users is affected by user mobility, and changes in channel conditions lead to differences in signal-to-noise ratios (SNRs) between users. Therefore, an independent symbol rate needs to be allocated to each user to ensure reliable communication, where the target bit error rate (BER) must be ensured to be below 10⁻³. The resulting channel capacity is denoted as r. i (n), i=1,2,3….

[0037] For user services, this embodiment divides the requirements into two basic dimensions: first, the long-term high throughput and medium latency requirements of file transfer (FT) services; and second, the low latency and medium throughput requirements within a short scheduling cycle of real-time interaction (RTI) services. For example, users performing file transfers expect the transfer to complete within seconds, while real-time interaction users need to continuously receive data packets per second to maintain real-time responsiveness. These service requirements can be quantified into the specific throughput needed by each user, using... This indicates that the period length is T. c Under a fixed-period scheduling, assuming the time slot allocated to each user is [duration missing] The remaining time slots for scheduling are denoted as T. i re (n), where n represents the index of the current scheduling period.

[0038] Reference Figure 3The diagram shows a block diagram of an indoor multi-user downlink wireless optical communication network system. The system consists of two main parts: a transmitter and a receiver. The transmitter is fixed, while the receiver's location is random but stationary. The transmitter includes a source module, a central control module, and a wireless light source. The central control module further includes a signal modulation module, a time slot scheduling module, a light source driving module, and a feedback receiving module. The optical signal transmitted by the transmitter is transmitted to the receiving user through an indoor spatial channel. The receiver consists of a signal acquisition module, a signal demodulation module, a data volume statistics module, an uplink feedback module, and a sink module. At the transmitter, the source module generates the data to be transmitted, such as files to be downloaded. This data is then passed to the central control module, where the signal modulation module modulates the data from the source module to reduce data transmission interference. The feedback receiving module receives location information, demand information, and service type information from the user end and sends this feedback information to the signal modulation module and the time slot scheduling module. The time-slot scheduling module first calculates the channel capacity for each user, then calculates the total number of time slots allocated to each user based on the feedback demand information, reconstructs a suitable data frame structure according to the service type, and sets the corresponding user's transmission symbol rate. After passing through the time-slot scheduling module, the time-domain signal determining when to send which signal is generated. The light source driving module receives the signal, loads it onto the wireless light source, and then transmits the optical signal through the downlink of the room. At the receiving end, after the signal acquisition module receives the optical signal, the signal demodulation module uses simple filtering, extracts and demodulates the processed signal, and outputs it to the data volume statistics module. The data volume statistics module counts the amount of valid data transmitted and calculates the remaining value of the required throughput based on the transmission time, sending it to the uplink feedback module. The uplink feedback module at the receiving end and the feedback receiving module at the transmitting end form a feedback channel. The feedback channel is a generalized channel and is not limited to one type; it can be a wireless Wi-Fi channel or a wireless optical channel. The sink module performs the final signal recovery and performance analysis.

[0039] Based on such Figure 2 and Figure 3 The indoor multi-user downlink wireless optical communication network and system shown herein, running the hybrid fair scheduling method provided by this invention, specifically includes: In the SORR (Service-Oriented Round Robin) scheduling algorithm that considers the throughput requirements of individual users, the ratio of the service throughput required by a user to its channel capacity is defined as the real-time demand slot, expressed as: ; in, This can be understood as meeting user throughput requirements. The required time; a larger demand coefficient indicates that users need to occupy more time slot resources to meet their needs; therefore, based on The weighting coefficients are calculated to ensure proportional allocation while preventing scheduling cycle overflow. Then... Normalization is performed to obtain the normalized demand factor: ;in, Total number of users; Subsequently, based on this normalized demand factor, the amount of data that can be allocated to users within the current scheduling period is calculated. The maximum time slot length is expressed as: ; Regardless of how demand factors influence the weighting process, a smaller value is typically chosen when determining the final time period allocation. This ensures that any excessive throughput demand will be truncated, and that the allocated time slots are sufficient to meet (but will not exceed) the actual service needs of each user.

[0040] After determining the time allocation using the SORR algorithm, the PF algorithm is adopted to optimize the time slot allocation within the period to reduce service latency. In the PF scheme, the first... Within the scheduling cycle, the first Scheduling priority for individual users Defined as: ; in, This indicates the number of times up to the current scheduling slot. Average throughput per user, duration is , The weighting factor is set to 0.95; To reduce transmission latency for Real-Time Interactive Users (RTIs), this embodiment proposes a Service-Oriented Weighted Scheduling Algorithm (SOHF). This algorithm adjusts scheduling priorities by introducing a service-oriented weighting factor to obtain the service-oriented priority. The service-oriented weighting factor is defined as the user's real-time demand slot. The ratio of the user's demand to the global average demand factor is logarithmically transformed to enhance the priority of high-demand users while avoiding excessive bias towards any one user. As user throughput demands are gradually met, their priority naturally decreases, thus prioritizing scheduling users with higher but unmet demands. This normalization mechanism ensures fairness and comparability among users, ultimately limiting the weighting factor to the interval [1, 1 + log(U)], and monotonically increasing with user demand. The service-oriented priority in the SOHF scheme is expressed as: ; in, Indicates user During the scheduling period The initial throughput needs to be noted. The time slot allocation will gradually decrease as user demand is met, and this adjustment mechanism ensures fairness when allocating time slots to RTI users in each scheduling cycle.

[0041] Based on the above description, this embodiment provides an implementation process for the service-oriented weighted scheduling algorithm SOHF, specifically including: Input: Total number of users FT user numbers and the number of RTI users User throughput demand matrix User channel capacity matrix reallocate minimum time slot and cycle .

[0042] initialization: set up , And set the allocation time slot matrix. and the actual transmission queue ; Set the average user demand FT user matrix and RTI user matrix ; when At that time, the while loop is executed: (1) Generate the service type matrix: If the user service type is RTI: set up Otherwise set ; (2) Generating demand factors: Demand Factors and generating the initial demand factor ; (3) Setting ; End the while loop: (4) Calculate the demand factor The sum, and the result of the summation is expressed in letters. Indicate; calculate the initial demand factor The sum, and the summation result is used It means that, among them, ; when At that time, the while loop is executed: (5) Generate normalization factor: Generate normalization factor and initial normalization factor ; Calculate the maximum allocation slot , ; calculate ; set up ; (6) Setting , ; End the while loop For any Execute the for loop: (7) Truncation restrictions for computation time: if : set up

[0043] End the for loop (8) The extra time that is truncated and shared equally by the remaining users. ; when ,and Execute the while loop: (9) Perform time slot reorganization based on the time slot lengths already allocated to users as described above: For any Executing a for loop: Calculate priority ; End the for loop Find the user with the highest priority and record its sequence number. Then set And update ; End the while loop (10) Record the number of remaining unreassembled time slots: For any Execute the for loop: Calculate the number of remaining unrecombined time slots. Updated later ,in The number of cycles is indivual; End the for loop Output: , .

[0044] This invention proposes a Service-Oriented Hybrid Fairnessscheduling (SOHF) method, aiming to better meet the service needs of multiple users with similar service quality levels while maintaining fairness and reducing average latency. The algorithm dynamically allocates time slots by comprehensively considering user channel capacity and throughput requirements, and reconstructs the scheduling structure according to different service types. The key point is that this method strives to provide all users with similar throughput, unaffected by differences in channel capacity, and significantly reduces waiting latency while achieving better fairness.

[0045] Based on the above embodiments, in this embodiment, the hybrid fair scheduling method provided by the present invention is used to schedule the system shown in Table 1, i.e., the downlink wireless optical communication network in a 5m×5m×3m room includes two file transfer users (user 1 and user 2) and one real-time interactive user (user 3). The light source emits 20W of optical power in Lambertian radiation mode. Unless otherwise stated for comparative purposes, the user positions are fixed as: user 1 (0,0,0.85), user 2 (1,1,0.85), and user 3 (2.5,2.5,0.85). The downlink uses on-off keying (OOK) modulation. At the above positions, the channel capacities required to achieve the target bit error rate of 10⁻³ are: r1=14 Mb / s, r2=31 Mb / s, and r3=51 Mb / s, respectively.

[0046] Table 1 System Parameters Room size 5m×5m×3m Light source position (2.5,2.5,3) User 1 location (0,0,0.85) User 2 location (1,1,0.85) User 3 location (2.5,2.5,0.85) Light source transmission power 20 W Modulation format OOK Modulation index 0.2 Light source half-power angle 60 deg Room receiving floor height 0.85 m PD field of view 85 deg PD receiving area <![CDATA[10 −4 m 2 ]]> PD Response 0.35 A / W

[0047] The system is managed by a central controller that generates frame data and sends it to the wireless light source for transmission. Frame arrival follows a Poisson process, with all users experiencing the same arrival rate. Each frame is 1000 bytes in size, and inter-frame guard intervals are ignored. The total simulation duration is 100 seconds. In the PF algorithm, the time frame is divided into 1000 time slots per second, each lasting 1 millisecond. Similarly, the time interval in the SOHF algorithm... It is also set to 1 millisecond.

[0048] Demand load (L) is defined to reflect the ratio of each user's throughput request to its average channel capacity. Its calculation formula is: ;in, In the simulation, all users are assumed to be identical, i.e., to conduct fair communication.

[0049] The formula for calculating Jain's Fairness Index (JFI) is as follows: ;

[0050] Specifically, the hybrid fair scheduling process includes:

[0051] Step 1: Traverse the load L, taking values ​​from 0.1 to 0.9.

[0052] Step 2: Since the user's location is known and fixed, we can obtain... That is, the average channel capacity is (14+31+51) / 3=32Mb / s, and it will not change with... The changes will occur accordingly. At this point, both User 1 and User 2 can adjust their needs based on the required values. Calculate This value represents the magnitude of the demand factor, which is also the actual length of the time slots allocated to a user to fulfill their demand throughput. When the product of the user's normalized coefficient and the polling cycle length exceeds the actual demand length, time slots are allocated according to the actual demand. Otherwise, the maximum is the size of this product value, and any excess will be truncated.

[0053] Step 3: Even if user needs change, assuming one user has a greater transmission demand, resulting in a larger required throughput; if the weighted product calculated from the increased throughput or the actual required length is greater than before the throughput change, the allocated value will still be limited. This avoids over-allocation; and the threshold value for this cutoff is strongly correlated with the user's channel conditions.

[0054] Step 4: After obtaining the user's time slot allocation value, allocate it according to the time interval. The size is broken down into several quantities, and the calculation result is: In this embodiment, the number of RTI users is used for time slot reallocation, with each... Calculate the service orientation priority for each user based on the size of the time slot. The time slot will be allocated to the highest priority user. Generally, the first reallocation will select the user with slightly better channel conditions and higher throughput demand. In this example, user 3 will be prioritized. When user 3... After the allocation of time slots is completed, since users 1 and 2 still have time slots and their business requirement is file transfer, the reallocation will end, and the remaining time slots will be superimposed and transmitted in a concentrated manner to ensure stability and reduce the latency waiting time of context switching.

[0055] Step 5: At this point, we will obtain the size of the time slot allocated to a user in each cycle and an actual transmission queue. This manifests as user 3 alternating with users 1 and 2 to transmit data according to priority, followed by user 2 transmitting files in a concentrated manner, and then user 3 transmitting files in a concentrated manner.

[0056] by Taking numerical calculations as an example, when users seek fairness, then... , , , After normalization , , ,but , , JFI=1; RTI users (i.e., user 3) have 104 time slots, while FT users (i.e., user 1 and user 2) have 381 and 172 time slots respectively. Priority is then calculated for these 104 time slots. If users have different demand levels in the first step, for example, a 2:1:1 ratio, then... , , , , , 078. After normalization , , 00, then , 29, 078; JFI=0.98.

[0057] Reference Figure 4 The figure shows a comparison of the joint fairness index of various scheduling algorithms under different user demand loads. The PF algorithm exhibits lower fairness due to its simultaneous pursuit of fairness and throughput balance, while the RR algorithm achieves higher fairness. The SORR algorithm can completely guarantee user fairness (JFI=1), and even after time slot reorganization, the fairness of the SOHF algorithm remains unchanged. Furthermore, the increase in demand load does not affect the fairness among users because once a user's location is fixed, their channel capacity is also fixed. Therefore, as the demand load increases, the throughput demand of each user... Growing at the same rate leads to time slot allocation Keeping it unchanged will keep the JFI value constant.

[0058] Reference Figure 5 The figure shows a comparison of the joint fairness index of various scheduling algorithms under different user channel capacities; the users are distributed in multiple different locations, and their capacity combinations are labeled as ( , , As user capacity differences increase, the fairness of both the RR and PF algorithms decreases to varying degrees, with the PF algorithm showing particularly significant fairness degradation when capacity differences are substantial. In contrast, the SORR and SOHF algorithms maintain a significant improvement in fairness under these conditions, keeping the JFI value stable at 1.

[0059] Reference Figure 6The figure shows a comparison of user throughput for various scheduling algorithms under different loads, illustrating the actual throughput achieved by users as system load increases, including the total throughput of all users and the throughput of the user with the worst channel conditions (User 1). The overall throughput of the SORR and SOHF algorithms is slightly lower than that of the traditional PF and RR algorithms. However, when the load exceeds L>0.8, User 1's throughput begins to surpass that of the PF and RR algorithms, with a 4.5% improvement compared to RR and a 74.3% improvement compared to PF. As the load continues to climb, the SORR and SOHF algorithms cannot allocate more throughput to User 1, causing its throughput to plateau. This improvement is achieved at the expense of users with better channel conditions, but it ensures overall fairness.

[0060] Reference Figure 7 The figure shows a comparison of waiting latency for different users under various scheduling algorithms when the demand load is 0.5. It compares the waiting latency experienced by different users under each scheduling algorithm. This embodiment assumes that the propagation latency is negligible and the processing latency is fixed at 0.01 milliseconds; Figure 7 It can be seen that the waiting latency achieved by the PF algorithm is lower than that of the RR and SORR algorithms, while the latency of SORR and RR is similar. This is because the Fair Priority scheduling algorithm allocates more time slots to users with poor channel conditions. When the demand load L=0.5, under the condition of meeting the required throughput, the waiting latency of each algorithm is basically the same. For RTI users (user 3), the SORR algorithm shows little improvement in waiting latency compared to the RR algorithm, while the SOHF algorithm achieves the lowest latency. Therefore, under the premise of maintaining equal fairness, SOHF can effectively reduce latency compared to SORR.

[0061] Reference Figure 8The figure shows a comparison of the average latency of various scheduling algorithms under different demand loads, comparing the average latency under different load conditions. The PF algorithm achieves the lowest average latency because even under the worst transmission conditions (i.e., the packet arrives just as the user's time slot ends), the waiting time will not exceed one short polling cycle; the packet can be scheduled immediately in the next cycle, at which point the user priority reaches its maximum value. In contrast, under similar conditions, the RR algorithm may need to wait for one or more longer polling cycles. The SOHF algorithm inherits the advantages of PF and further optimizes the latency for users with RTI service requirements. When L=0.6, the average latencies of PF, SOHF, SORR, and RR are 0.86 ms, 0.88 ms, 0.93 ms, and 0.97 ms, respectively. The SOHF latency is close to that of PF and 5.4% lower than SORR. SORR reduces latency by 4.1% compared to RR. When L < 0.5, SORR performs similarly to RR; when L > 0.5, SORR alleviates the congestion problem of RR by allocating longer time slots to users with poor channel conditions, thereby delaying the sharp rise in average delay to around L ≈ 0.6, while maintaining stable performance under moderate load.

[0062] The method proposed in this invention integrates the advantages of RR and PF scheduling and employs an optimization mechanism based on required throughput to allocate user time slots, improving fairness while maintaining low waiting latency. Under fairness requirements, this embodiment achieves a joint fairness index of 1, solving the problem of users being unable to communicate fairly due to objective communication limitations, and features low complexity and high adaptability. Compared to the RR scheme, the proposed method can improve throughput slightly for users with poor channel conditions, but compared to the PF scheme, it provides a more significant improvement for users with poor channel conditions. In terms of latency performance, SOHF achieves the lowest latency for RTI users, showing a significant improvement over the RR scheme, particularly in multi-user scenarios where fairness is a critical requirement.

[0063] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0064] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A hybrid proportional fair scheduling method based on a wireless optical communication network, characterized in that, Comprise: S1: based on the delay sensitivity of each user's to-be-transmitted data and throughput demand, all users are divided into transmission type users and interactive type users; S2: real-time acquisition of all users in the current scheduling period, real-time demand throughput and real-time channel capacity in wireless optical communication network; S3: for each user, the ratio of real-time demand throughput and real-time channel capacity is calculated, the real-time demand time slot is obtained, and the normalization is carried out to obtain the normalized demand factor; based on the product of the normalized demand factor and the preset cycle period, the maximum allocatable time slot is obtained; the smaller value of the maximum allocatable time slot and the real-time demand time slot is selected as the actual allocation time slot of each user; S4: the ratio of the actual allocation time slot of each user and the preset reallocation minimum time slot is obtained as the to-be-recombined time slot number of each user; S5: the average value of real-time throughput demand of each user in the historical cycle period is calculated as the historical average throughput, combined with the real-time channel capacity, to obtain the initial priority of each user; S6: based on the real-time channel capacity of each user and the initial demand throughput of the current scheduling period, the initial demand time slot of each user is constructed, and the average is obtained to obtain the global average demand factor of the current scheduling period; S7: the ratio of the real-time demand time slot of each user and the global average demand factor of the current scheduling period is logarithmically transformed, multiplied by the initial priority of each user, to obtain the service-oriented priority of each user; S8: the interactive type user with the maximum service-oriented priority is obtained, and a time slot is allocated to the interactive type user, and the to-be-recombined time slot number of the interactive type user is reduced by one; S9: return to step S2 until the to-be-recombined time slot number of all interactive type users is 0, and the time slots are allocated to the transmission type users in turn according to the service-oriented priority from high to low.

2. The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 1, wherein, For each user, the ratio of real-time demand throughput and real-time channel capacity is calculated, the real-time demand time slot is obtained, and the normalization is carried out to obtain the normalized demand factor, comprising: computing real-time demand throughput for each user with real-time channel capacity , real-time demand slots for each user are obtained , expressed as: ; The real-time demand time slots of each user are normalized to obtain normalized demand factors of each user , is expressed as: ; wherein, denotes the demand throughput of the th user in the th scheduling period, denotes the channel capacity of the th user in the th scheduling period, denotes the normalized demand factor of the th user in the th scheduling period; , denotes the total number of users.

3. The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 2, wherein, The smaller value of the maximum allocatable time slot of each user and the real-time demand time slot is selected as the actual allocation time slot of each user, represented as: ； wherein, represents the actual allocated time slot of the th user in the th scheduling cycle; represents the maximum allocatable time slot of the th user in the th scheduling cycle, expressed as , represents a preset cycle period.

4. The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 1, wherein, The average value of real-time throughput demand of each user in the historical cycle period is calculated as the historical average throughput, combined with the real-time channel capacity, to obtain the initial priority of each user, represented as: ； wherein, represents an initial priority of the th user in the th scheduling cycle, represents a real-time channel capacity of the th user in the th scheduling cycle, represents an average of real-time throughput demand of the th user in a history cycle period of the th scheduling cycle; represents a preset weight factor.

5. The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 4, wherein, Based on the real-time channel capacity of each user and the initial demand throughput of the current scheduling period, the initial demand time slot of each user is constructed, and the average is obtained to obtain the global average demand factor of the current scheduling period, comprising: based on real-time channel capacity of each user initial demand throughput of the current scheduling period , the initial demand time slot of each user is constructed, denoted as: ; The average of the initial demand slots of all users is calculated as the global average demand factor for the current scheduling period, denoted as: ; wherein the initial demand throughput of the first user in the first scheduling period is the real-time demand throughput of the first user when the first user is allocated the first time slot in the first scheduling period. wherein the initial demand throughput of the first user in the first scheduling period is the real-time demand throughput of the first user when the first user is allocated the first time slot in the first scheduling period.​​​​ 6. The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 5, wherein, The ratio of the real-time demand time slot of each user and the global average demand factor of the current scheduling period is logarithmically transformed, multiplied by the initial priority of each user, to obtain the service-oriented priority of each user, represented as: ； wherein, represents the service-oriented priority of the th user in the th dispatch cycle. 7.The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 1, wherein, The time slots are allocated to the transmission type users in turn according to the service-oriented priority from high to low, comprising: calculating the to-be-recombined time slot number of each transmission type user, and allocating the corresponding to-be-recombined time slot number of each transmission type user to the transmission type user in turn according to the service-oriented priority from high to low. 8.The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 1, wherein, The to-be-recombined time slot number of each transmission type user is calculated, comprising: Obtaining the remaining time slots to be allocated to each of the transmission type users after the number of time slots to be reorganized of all the interactive type users is 0 in the current scheduling period; Calculating the ratio of the remaining time slots to be allocated to each of the transmission type users to the preset minimum time slot for re-allocation as the number of time slots to be reorganized of each of the transmission type users. 9.The hybrid proportional fair scheduling method based on wireless optical communication network according to claim 1, wherein, After the time slots to be allocated to all the transmission type users are ended, including: In the current scheduling period, constructing an initial transmission queue based on the time sequence in which all the interactive type users are allocated the first time slot; Adding all the transmission type users to the initial transmission queue in the order from high to low based on the service-oriented priority to obtain an actual transmission queue; Obtaining the actual allocated time slots of each of the interactive type users and the transmission type users to form a time slot matrix.

10. A hybrid proportional fair scheduling system based on a wireless optical communication network, characterized in that, Including: Integrated in the transmitting end: A source module for obtaining the data to be transmitted of each user; A central control module for obtaining the actual transmission queue and the time slot matrix of all the users for transmission based on the hybrid fair scheduling method based on the wireless optical communication network according to any one of claims 1 to 9; An optical source driving module for generating a driving signal based on the actual transmission queue and the time slot matrix of all the users for transmission; A wireless optical source for emitting corresponding optical signals based on the driving signal; Each user of the receiving end for demodulating and extracting based on the optical signals to obtain the transmission result of the corresponding data to be transmitted.

Images