Layered Coded Cache Placement and Cooperative Delivery with Sharing Links in Satellite-Terrestrial Integrated Networks

Gu Shushi ,Chen Zihan ,Wu Yaonan ,Zhang Qinyu,2, ,Wang Ye

1 School of Electronic and Information Engineering,Harbin Institute of Technology(Shenzhen),Shenzhen 518055,China

2 Guangdong Provincial Key Laboratory of Aerospace Communication and Networking Technology,Shenzhen 518055,China

3 Broadband Communications Research Department,Pengcheng Laboratory,Shenzhen 518052,China

Abstract: Cooperative utilization of multidimensional resources including cache,power and spectrum in satellite-terrestrial integrated networks (STINs) can provide a feasible approach for massive streaming media content delivery over the seamless global coverage area.However,the on-board supportable resources of a single satellite are extremely limited and lack of interaction with others.In this paper,we design a network model with two-layered cache deployment,i.e.,satellite layer and ground base station layer,and two types of sharing links,i.e.,terrestrial-satellite sharing(TSS) links and inter-satellite sharing (ISS) links,to enhance the capability of cooperative delivery over STINs.Thus,we use rateless codes for the content divided-packet transmission,and derive the total energy efficiency (EE) in the whole transmission procedure,which is defined as the ratio of traffic offloading and energy consumption.We formulate two optimization problems about maximizing EE in different sharing scenarios (only TSS and TSS-ISS),and propose two optimized algorithms to obtain the optimal content placement matrixes,respectively.Simulation results demonstrate that,enabling sharing links with optimized cache placement have more than 2 times improvement of EE performance than other traditional placement schemes.Particularly,TSS-ISS schemes have the higher EE performance than only TSS schemes under the conditions of enough number of satellites and smaller inter-satellite distances.

Keywords: coded content placement;cooperative delivery;energy efficiency;sharing links;STINs

I.INTRODUCTION

The explosive growth of popular mobile data traffic with limited capacity of backhaul links has brought tremendous pressure to the traditional terrestrial cellular networks to satisfy users’ quality or experience requirements towards 6G networks.A variety of content caused by on-demand multimedia exhibits reusing characteristics according to the users’preferences.Using mobile edge caching (MEC) technology,popular content can be cached at the edge of the wireless network,avoiding frequent fetching from the core cloud,reducing the delivery delay and saving the backhaul bandwidth.The future wireless network with MEC will deploy dense small cells with caching capability over the hierarchical macro and micro cellular infrastructures[1].

However,the ground backhaul network uses multihop unicast,thereby the cached content must pass through multiple hops to each base station separately,to offer ubiquitous connectivity.Fortunately,satellite system can provide a broadband backhaul link covering a wide area [2],and broadcast content to all base stations and users in anytime.Therefore,the combination of MEC and satellite communication to deal popular content distribution can further alleviate the load pressure of the core network,and improve broadband communication qualities of services (QoS).Recently,the satellite internet and the large-scale low earth orbit(LEO) satellite constellation are emerged to provide global internet access.Meanwhile,satellite-terrestrial integrated network(STIN)is gradually confirmed as a new development trend to the 6G networks.

The study of caching-assisted content delivery in STIN remains as a hot topic recently.The authors of [3] propose an STIN cooperative transmission scheme to offloading traffic from base stations,so as to realize an energy-efficient radio access network(RAN).And[4]proposes a novel three-layer satellite network to provide flexible management and efficient content retrieval for STINs.In[5],the scenario where multiple satellites jointly serve multiple ground terminals is considered,and the content placement problem is solved by using a matching algorithm based on LEO satellite constellation.[6] and [7] both consider the caching resource optimization in the spaceair-ground integrated network (SAGIN).[8] and [9]investigate the beamforming in STIN,and jointly optimize the cache placement,AP clustering,and multicast beamforming by decomposing the optimization problem into a short-term subproblem and a long-term subproblem.

Nevertheless,in the existing STIN,the cache and energy resources are both limited on board[10].Thus,there are many researches focusing on the sharing scheme,such as inter-satellite sharing links,to solve resource-constrained problems.[11] investigates the cooperative multi-layer edge caching in STIN,which includes cooperative caching of satellites,and formulates the content placement problem to minimize the average content retrieving delay of users.[12] proposes a load balancing routing algorithm for LEO constellation network based on extended link state to reduce link congestion and packet loss rate.In respect of heterogeneous network(HetNets),the authors in[13]provide two potential structures for 6G communications to solve the resource allocation (RA) problems of the HetNets,i.e.,a learning-based RA structure and a control-based RA structure.Based on energy efficiency (EE),the authors in [14] integrate reconfigurable intelligent surfaces (RISs) into the heterogeneous network (HetNet) and study a realistic robust beamforming algorithm with max-min fairness for an RIS-aided HetNet under channel uncertainties and hardware impairments (HWIs).And [15] considers the changed rule and cache update strategy of inter-satellite link in STINs,and proposes a densitybased network division algorithm to improve content distribution efficiency.In[16],the multi-layered satellite network (MLSN) is considered to extend storage resources of LEO,and the authors propose a cache scheme based on the Stackelberg game.However,it is noted that layered architecture and sharing transmission will decrease the EE,which has not been considered by most studies.

It is well-known that,coded caching has arisen as a promising solution to reduce the downlink traffic[17],and provided remarkable improvements in content delivery time [18,19],dividing each file into multiple segments and strategically caching different segments of the file at edge nodes.[20]and[21]both optimize the coded cache placement in cache-enabled SBSs to improve communication performance.The work in [22] considers the cooperative transmission based on the maximum distance separable (MDS) codes in hierarchical industrial internet of things networks,and finds the optimal content placement for maximizing EE.Rateless codes are similar to MDS codes,but they have lower coding redundancy,where the receiver can recover files with a higher decoding success probability.It is very suitable for multimedia transmission with limited delay on fading channels in wireless communication systems,such as broadcast,relay and cognition,and also suitable for storage,compression coding and network transmission [23,24].Therefore,utilizing rateless coded caching in STIN to realize the cooperative caching and transmission,can further improve the EE.

Motivated thereby,in this paper,we consider a two-layered STIN model,including satellite layer and ground small base stations (SBSs) layer,which both can cache the rateless coded content packets.Moreover,terrestrial-satellite sharing (TSS) link and intersatellite sharing(ISS)link are used to further alleviate the pressure of limited cache resources on-board by cooperative delivery among the two layers.Then the optimal content placement strategy is elaborated in detail to maximize the EE under the different scenarios.The contributions of our work can be summarized as follows:

1) We propose an STIN with layered coded cache placement,which include satellite layer caching and SBS layer caching.Considering the limited cache capacity of satellite and the property of the rateless codes,we also propose the content sharing delivery schemes that the coded packets can be transmitted to satellite by TSS link and ISS link to satisfy the other SBSs’user preferences.

2) For the only TSS scenario,we derive the expressions of traffic offloading and energy consumption,and formulate a maximum optimization problem of EE to obtain the optimal content placement matrix under the constraints of cache size and power.And we propose TSS iterative cache placement(TSS-ICP)algorithm to jointly optimizing the content placement vectors of SBSs and LEO satellite,respectively.Simulation results show that,compared with most popular caching and share-disabled strategies,TSS-ICP can achieve several times EE improvement.

3) Under the TSS-ISS scenario including TSS link and ISS link at the same time,we formulate the maximum problem of EE,and propose a TSS-ISS genetic cache placement (TSS-ISS-GCP) algorithm for the feasible solutions.Simulation results demonstrate that,TSS-ISS-GCP has the better EE performance compared with the popular first caching schemes and the only TSS sharing schemes.

The rest of the paper is organized as follows: Section II describes the system model.In Section III,the expressions of traffic offloading and energy consumption are derived under two scenarios,and the EE maximization problems are also formulated.Section IV designs two algorithms to solve the two optimization problems for the optimal placement matrixes,respectively.Section V gives the numerical simulation results and analysis.Section VI concludes this paper.

II.SYSTEM MODEL

2.1 Network Model

As Figure 1 shown,an STIN system model is composed of four key components: LEO satellites,SBSs,macro base stations(MBS)and users.Each SBS covers a cell while multiple LEO satellites at the same orbital altitude periodically covers all cells.Only one LEO communicates with MBS and users,called the main satellite LEO 0,and the rest are auxiliary satellites,assuming that only adjacent satellites can communicate between each other with ISS links.The set of LEO satellites is denoted asL≜{0,1,2,...,L}.There areLauxiliary satellites in total,andl=0 stands for the main satellite.We denote the set of SBSs asN≜{1,2,...,N},and the associated users set asUn.SBSncan only serve users within the coverage area.Users are distributed on a homogeneous poisson point processes(HPPP)with densityθ0.

Figure 1.Cooperative content delivery in STIN system model with layered caches and sharing links.

The set of files is denoted asF{f1,f2,...,fI}.Without loss of generality,we assume that all files have the same size ofs.There is layered caching placement,which includes satellite layer and SBS layer equipped with finite storage capacityMSandMT,respectively.Satellites and SBSs serve users cooperatively by broadcasting the cached content.For MBS,we assume it is connected to core network via reliable fiber backhaul,and will serve users through backhaul link if content retrieval failure happens.

2.2 Rateless Coded Caching Model

We use a rateless coded caching scheme,i.e.,Lubytransform(LT)codes,to encode the message at packet level.LT codes allow successful recovery of original files with very a small overhead.Each files is divided intokoriginal packets with the same size ofz=Then we selectd(1≤d ≤k)original packets independently and randomly for bitwise exclusive-or(XOR),which can obtain infinite coded packets.Users can recover original file from the received arbitraryK=k(1+ε)(Kis slightly larger thank)coded packets with high decoding success probability.It can be seen that,rateless code can ensure the content cached in SBSs and satellite are totally different,which obtains higher caching efficiency.We denote the number of coded packets cached in SBSnand LEOlasandrespectively.In this STIN system,we can improve the delivery EE by changing coded packets placement strategy.

2.3 File Request and Delivery Model

We assume that the file popularity follows the Zipf distribution[20].The probability of a user in SBSnasking fori-th file is described as

whereγi,nis the popularity rank ofi-th file in SBSnandαis skewness factor.We assume that the files popularity is heterogeneous in SBS cells by setting different popularity ranks.

The user in the coverage of SBSnwill send a request fori-th file to MBS,then MBS will schedule satellite and SBSntransmit cached packets within a small time slott.We assume that SBSs and satellite can simultaneously send the content to users on nonoverlapping frequency bands.User request can be satisfied by two manners,only SBSnbroadcasting,or satellite and SBSntransmitting content cooperatively,i.e.,satellite broadcasts the rest of packets if SBSndoes not cache enough coded packets for the requests.

We also consider a cache sharing manner in the STIN system.There two delivery sharing links,i.e.,TSS link and ISS link.In detail,if users do not get enough packets from SBSnand the main LEO satellite,SBSs in other cells can share the cached packets to the main satellite through MBS,or auxiliary satellites share packets to the main one.We introduce the share variableai.When the value is 0,it means that only the TSS link is enabled and the cache placement matrix in the auxiliary satellite is cleared.Whenaiis valued at 1,it means that only the ISS link is enabled and the cache placement matrix in SBSs is cleared.

2.4 Channel Model

2.4.1 Delivery Links Delivery link include satellite-to-user link and SBSto-user link.In the downlink of satellite,we mainly consider the rain attenuation following Weibull distribution.The power attenuation of the satellite-to-user link can be

whereλis the wavelength of carrier,Frainis rain attenuation,GTXandGRXare antenna gains of the satellite and users devices respectively,andHSis the height of LEO satellite.The maximum transmission rate is

whereWSis satellite communication bandwidth,||σS||2is the noise power,andis the transmission power of satellite.To ensure satellite can transmitpackets within time slott,the transmission rate is expressed as

While the transmission power of satellite is derived by

The large-scale fading and small-scale fading are both considered in SBS-to-user link.The maximum transmission rate of SBS-to-user link is computed by

whereWTis SBSs communication bandwidth,rnis the average distance between SBSnand users,βis the path loss factor,h～N(0,1)is fading coefficient,and||σT||2is the noise power.So,the transmission power of SBS-to-user link is

2.4.2 Terrestrial-Satellite Sharing Links

In TSS links,the requested content will be delivered from SBSs to the satellite through MBS.And we mainly consider the SBS-to-MBS link,in which exists large scale fading[3].Similar to the delivery link,the transmission power of SBS-to-user link and SBS-to-MBS link is computed by

2.4.3 Inter-Satellite Sharing Links

We only considers free space attenuation in ISS link,and the transmission power can be expressed as

where||hISL||2=is fading coefficient of the inter-satellite link,andHISLis the distance of the adjacent satellites.

III.ENERGY EFFICIENCY MAXIMIZATION PROBLEM FORMULATION

3.1 Scenario 1: Only Terrestrial-Satellite Sharing

In the only TSS scenario,cache sharing inter satellites is disabled.First,the cache matrix of the auxiliary satellite is set to 0,that is,only the TSS link is considered.When the sum of SBSnand main satellite cache coding packets is less thanK,the ground sharing link is enabled.

3.1.1 Traffic Offloading

The traffic offloading from SBSs is expressed as

whereUi,n=|Un|µi,ndenotes the average users that requesti-th file SBSncoverage area,and|Un|denotes the number of users in SBSn.The traffic offloading served by satellite is expressed as

where minn∈Ndenotes the minimum number of packets cached in SBSn.

3.1.2 Energy Consumption For EE in STIN,we compute the energy consumed by both SBSs and the satellite.Since SBSs can provide multiple access to each user and cache sharing to the satellite,the energy consumption of SBSs’ terrestrial transmission can be computed by

The energy consumption of LEO satellite can be computed by

3.1.3 Energy Efficiency Maximization

We denote the EE of entire system as the ratio of the total traffic offloading and energy consumption,which can be respectively computed as

Thus we can formulate a maximizing EE optimization problem,which can be expressed as

3.2 Scenario 2: Terrestrial-Satellite Sharing and Inter-Satellite Sharing

In the TSS-ISS scenario,TSS link and ISS link both exist.The traffic offloading is same as that in only TSS scenario,while the energy consumption of SBSs is derived by

The energy consumptions of the main satellite and the auxiliary satellites by ISS links are

The total traffic offloading and energy consumption can be respectively denoted as

Thus we can formulate a maximizing EE optimization problem,which can be expressed as

IV.LAYERED CACHE PLACEMENT OPTIMIZED ALGORITHM

4.1 Iterative Cache Placement Algorithm for Scenario 1

According to Eqs.(14)and(15),we can obtain the total traffic offloading in cellnand the total energy consumption for transmittingi-th file,respectively,i.e.,

To simplify the calculation,the original problem(16)can be transformed to

whereηis the optimal value of EE.

Then we propose a terrestrial-satellite sharing cache placement based on iterative algorithm(TSS-ICP)for the solution to the above problem as shown in Algorithm 1.

The basic idea of TSS-ICP is jointly optimizing the content placement vectors of SBSs and LEO satellites and updatingηby the iterative algorithm.Firstly,TSSICP algorithm maximizes the traffic offloading and gets the SBSncache placementand then,it works out the satellite cache placement by minimizing the energy consumption.Cache sharing is disable when=K-minn∈Nand then it adjustsfor the tradeoff of EE.We can analyze that the computational complexity of TSS-ICP isO（（MT+MS）·I·N）,which is a feasible solution for the content placement.

4.2 Genetic Cache Placement Algorithm for Scenario 2

Because of the cache variables involving multiple sharing links,double-layer and multiple nodes,we use Multiple sharing cache placement algorithm(TSSISS-GCP)based on the genetic algorithm as shown in Algorithm 2 to solve the cache placement matrix.

Firstly,andis solved according to the TSS-ICP algorithm,and we randomly generatenPopgroupsai.Then,under the constraints of cache capacity and power,SBSs and auxiliary satellites place the remaining required coded packets according to the file popularity,and generate the parent population withnPopindividuals.The fitness is the total EE of the system.The value ofaiof the individual with higher fitness is taken,and the number of coded packets in SBSs and satellite is adjusted according to the file popularity,which will be used to generate new individuals,update the population and export the fitness ranking individuals.We can analyze that the computational complexity of TSS-ISS-GCP isO（（（MT+MS）·N+GEN·nPop）·I）.

V.NUMERICAL SIMULATION RESULTS

5.1 Results of Scenario 1

We consider a scenario with one LEO satellite covering several cells,and the area of each cell is 1×1 km2.The distribution of users follows poisson point processes with same densityθ0.We assume the satellite and each SBS have the same cache size,which is denoted byThe other parameters are listed in Table 1.The values of parameters in this table refer to references[25]and[26].

Table 1.Parameters setting.

In this section,we simulate how the parameters(density of users,number of cells,cache size and Zipf parameter) effect the total EE in the proposed TSSICP algorithm and compare with the following benchmark algorithms.

•TSS-ICP:Considering only TSS link and using the joint content placement based on iterative algorithms.

•TSS-MPP: Only TSS link exists.SBSs and the satellite choose files to cache depending on the file popularity.

•NS-ICP:Using the joint content placement based on iterative algorithms,while cache share is disabled.

•NS-MPP:Popularity first and without considering cache sharing.

In Figure 2,we compare algorithms with different densities of users.We can find out that with the increase of density of users,EE of all strategies increases linearly.The reason is that SBSs and satellite both transmit content by broadcasting.Therefore,the increasing of users density has no effect on the power consumption,but traffic offloading will increase linearly.It can be seen that compared with TSS-MPP scheme,TSS-ICP can achieve 32.6%improvement of EE with the density of 60 users/km2.When the density of users is equal to 10,the EE of the TSS-MPP or NS-MPP is smaller than TSS-ICP or NS-ICP.This is due to the low density of users and the small number of requests,in which case content placement based on simple iterative algorithm performs better than the algorithm based on file popularity.Apparently,for the two nonshare strategies,the performance is worse than share-enable schemes.

Figure 2.EE versus the density of user θ0, with skewness factor α=1, the number of SBSs is 10 and the cache size is 10%.

And in Figure 3,we simulate the EE with respect to the skewness factorα.We can see that,by setting minimum cache size as 10%,TSS-ICP has maximum 2.84 times improvement compared with NS-MPP.The file requests are more centralized with larger skewness factorα.Therefore,with the growth ofα,the results of two schemes which cache content according to the request probability increase.It can be observed that the performance of TSS-ICP and TSS-MPP blue is better and more stable,because the cache sharing can achieve higher cache efficiency and take better advantage of the broadcasting of satellite.

Figure 3.EE versus the skewness factor α with the density of users θ0=100/km2,cache size is 10%and the number of SBSs is 10.

Figure 4 shows the relationship between the EE and the number of SBSs.We can find out that TSS-ICP has 2.04-4.14 times improvement compared with NSMPP scheme.Furthermore,the EE of share scheme increases faster than the nonshare scheme with the increase of number of SBSs.SBSs provide the sharing link which can improve the EE,thus the number of SBSs has greater impact on the algorithms with TSS link.

Figure 4.EE versus the number of SBSs, with the density of users θ0=100 /km2, cache size is 10% and skewness factor α=1.

Figure 5 shows the relationship between the total EE and cache size.It can be seen that TSS-ICP has better performance,which has 1.12-2.08 times improvement compared with NS-MPP scheme.Since both satellite and SBSs broadcast packets to users,the EE of all algorithms presents a decreasing trend,and the file popularity following Zipf distribution decreases exponentially.For the proposed TSS-ICP algorithm,larger cache size has less impact on total traffic than NS-ICP scheme,thus the total EE decreases faster.We can draw a conclusion that TSS-ICP has greater advantages especially when cache capability is very small.

Figure 5.EE versus the cache size,with the density of user θ0=100/km2,the number of SBSs is 10 and the skewness factor α=1.

5.2 Results of Scenario 2

We consider a scenario with multiple LEO satellites covering several cells in the same area.The area of each cell is 10×10 km2and the skewness factorα=1.The distribution of users follows poisson point processes with same densityθ0=100 users/km2.We assume each satellite and each SBS have the same cache size,which can be denoted byand,respectively.The other parameters are as same as those in Table 1.We simulate how the parameters (distance of inter-satellite,number of satellite,cache size of LEO and number of SBSs)effect the total EE,and compare the proposed TSS-ISS-GCP algorithm with TSS-ICP,TSS-MPP and the following two benchmark algorithms:

•TSS-ISS-GCP: Considering two share links and using the GA-based content placement.

•TSS-ISS-MPP: Considering two share links,and putting content depending on the file popularity.

In Figure 6,we compare different algorithms with different distances of inter-satellite.We can find out that with the increase of distance of inter-satellite,EE of TSS-ISS strategies decreases.That is beacause according to Eq.(9),the larger the inter-satellite distance is,the more power is consumed for transmitting coded packets,which results in the reduction of EE.The slight fluctuation of EE in TSS-ICP and TSS-MPP is caused by the accuracy of the ICP and MPP,but overall it fluctuates around an accurate value,which means the strategies with only TSS links are not affected by the distance of inter-satellite.And when the distance of inter-satellite is small,it can be seen that compared with TSS-MPP,TSS-ISS-GCP has 1.22-1.74 times improvement.

Figure 6.EE versus the distance of inter-satellite,with the number of satellites is 5,the cache size of LEO is 20%,the number of SBSs is 5 and the distance of terrestrial-satellite is 750 km.

In Figure 7,we simulate the total EE with respect to the distance of terrestrial-satellite.It can be seen that the EE of all algorithms presents a decreasing trend.And TSS-ISS-GCP has about 18.5%average improvement compared with TSS-ISS-MPP and achieves more than 2.31 times growth compared with TSS-MPP.Furthermore,the descent is similar for the all methods.Since auxiliary satellites do not communicate with the ground equipments,the satellite’s altitude does not affect the transmitted power.

Figure 7.EE versus the distance of terrestrial-satellite,with the distance of inter-satellite is 100 km, the number of satellites is 5,the cache size of LEO is 20%and the number of SBSs is 5.

Figure 8 shows the relationship between the total EE and the number of satellites.It can be found that the EE of TSS-ISS methods increases with the growth of the number of satellites and becomes stable when the number of satellites reaches a certain value.Since secondary satellites do not deliver coded packets directly to users,their efficiency does not increase as the number saturates.We can see that,by setting the number of satellites as 5,TSS-ISS-GCP has 11.2%-40.7%improvement compared with only TSS schemes.In addition,the number of satellites has no impact on the TSS link,which has better performance than TSS-ISS methods when the number of satellites is small.

Figure 8.EE versus the number of satellites, with the distance of inter-satellites is 100 km,the cache size of LEO is 20%,the number of SBSs is 5 and the distance of terrestrialsatellite is 750 km.

Figure 9 shows the relationship between the total EE and the cache size of LEO satellite.We can see that the EE of TSS-ISS has a trend of growth while the EE of the only TSS schemes presents a decreasing trend.Since the larger the LEO cache is,the weaker the role of satellite sharing plays,the file popularity following Zipf distribution decreases exponentially.It can be seen that,when the LEO cache is large,ICP algorithm has better performance in only TSS scenarios,which has 1.64-2.51 times improvement compared with TSS-MPP.

Figure 9.EE versus the cache size of LEO,with the number of satellites is 5,the distance of inter-satellite is 100km,the number of SBSs is 5 and the distance of terrestrial-satellite is 750 km..

In Figure 10,we simulate the total EE with respect to the number of SBSs.It can be seen that the EE of all algorithms presents a growing trend.And TSS-ISS-GCP has about 41.5%average improvement of EE compared with TSS-MPP.Furthermore,when the number of SBSs is small,TSS-ISS has slight growth compared with only TSS schemes,while GCP can achieve 36%improvement compared with popular first schemes.

Figure 10.EE versus the number of SBSs,with the distance of inter-Satellite is 100 km,the number of satellites is 5,the cache size of LEO is 20% and the distance of terrestrialsatellite is 750 km.

VI.CONCLUSION

In this paper,we consider an STIN system model with layered coded cache placement and propose the cache sharing cooperative delivery schemes,including TSS link and ISS links,to improve the system energy efficiency.Then we derive the expressions of traffic offloading and energy consumption,and formulate the maximum problems of EE under two sharing scenarios,i.e.,only TSS sharing scenario and TSS-ISS sharing scenario.Then we solve them by the designed iterative algorithm named TSS-ICP and genetic algorithm named TSS-ISS-GCP,respectively,to obtain the placement matrixes.Simulation results show that,TSS-ICP and TSS-ISS-GCP can raise significantly the EE for several times compared with other benchmark schemes.In addition,under the conditions of enough satellites and short distance of inter-satellite,TSS-ISS schemes have higher EE than only TSS schemes.In the future work,we will consider the cache updating strategy under multi-satellite cooperation scenario with several service periods.

ACKNOWLEDGEMENT

This work was supported by National Natural Sciences Foundation of China(No.62271165,62027802,61831008),the Guangdong Basic and Applied Basic Research Foundation (No.2023A1515030297,2021A1515011572),and Shenzhen Science and Technology Program ZDSYS20210623091808025,Stable Support Plan Program GXWD20231129102638002.