A joint resource allocation method for minimizing weighted delay energy

Abstract

A joint resource allocation method for minimizing weighted delay energy provided by the present invention comprehensively considers the NOMA user clustering strategy, the inter-cluster frequency band allocation in scheduling, and the power allocation within the control cluster. It groups users according to delay requirements and then forms ST units. Taking the allocated power of users as variables and minimizing the weighted delay energy as a constraint problem, it obtains the weighted energy of each ST unit under sub-channel resources. Then, through multiple iterations, the ST units are matched with the sub-channel resources. After the matching is completed, each user finds the most suitable resource allocation according to the delay requirements, achieving optimal resource allocation. Compared with the prior art, the present invention has the advantages of resource saving, high resource allocation efficiency, accurate resource allocation, and higher user experience.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and particularly relates to a joint resource allocation method for minimizing weighted delay energy. Background Art

[0002] After the fifth-generation mobile communication technology (5G) has seen substantial development, the rapid popularization of intelligent mobile devices has provided an interactive platform integrating various functions such as work, life, study, and entertainment. However, the explosion of data traffic and the diversification of computing types have brought unprecedented challenges to mobile devices with limited computing power and battery power. As an existing solution, Mobile Cloud Computing (MCC) adopts a centralized cloud computing center architecture. Users can migrate computing tasks to the cloud computing center through hierarchical network architectures such as radio access networks, backhaul networks, and core networks. After the cloud computing center executes the tasks, the results are then transmitted back to the users in the reverse direction. In this centralized computing network architecture, the long distance between the user and the cloud computing center is bound to result in long-distance core network transmission for computing task offloading, bringing high latency of hundreds of milliseconds. When the core network / backhaul network is congested, the latency can be as high as thousands of milliseconds. To overcome this problem, mobile edge computing sinks computing resources from the centralized cloud computing center to edge devices in the radio access network (such as base stations, routers, access points, etc.). The edge computing resources near the users provide computing services locally for the users, thereby avoiding the high latency generated by the backhaul network and the core network and reducing the task offloading latency to only dozens of milliseconds in the radio access network, significantly improving the latency experience of users' computing services.

[0003] With the continuous iteration of mobile communication technology, the new generation of B5G / 6G mobile communication services will cover new mobile communication services such as autonomous driving, intelligent industrial manufacturing, and remote medical treatment, giving rise to new requirements for millisecond-level ultra-low latency. However, currently, the radio access network faces a significant contradiction between "orthogonal access to limited spectrum resources and the continuously exploding number of users", making it difficult to provide millisecond-level task migration capabilities for a large number of users on limited spectrum resources. On the other hand, different from previous multi-user multiplexing multiple access technologies that mainly focus on the time domain, frequency domain, and code domain, NOMA technology adds a new dimension - the power domain. To achieve multi-user multiplexing in the power domain, a serial interference cancellation (SIC) module needs to be installed at the receiving end. Through this interference cancellation device, different users' signals can be distinguished at the receiving end. NOMA technology allows multiple users to use the same frequency band resource through multiplexing in the power domain, effectively alleviating the pressure on the frequency band resources and improving the system's latency performance. The development and use of NOMA technology will effectively improve the difficulties faced by MEC in terms of latency.

[0004] Due to the diversity and complexity of the energy consumption and latency requirements of access devices, in the MEC using NOMA, it is very necessary to perform scheduling for computing offloading, and of course, this task is very challenging and complex. In a typical multi-subchannel NOMA system, NOMA user pairing and subchannel allocation directly affect the achievable rate of NOMA users' uploaded tasks. Therefore, there is an urgent need for a resource allocation scheme for a hybrid NOMA-based MEC system. Summary of the Invention

[0005] To solve the above problems existing in the prior art, the present invention provides a joint resource allocation method for minimizing weighted delay energy. The technical problems to be solved by the present invention are achieved through the following technical solutions:

[0006] A joint resource allocation method for minimizing weighted delay energy provided by the present invention is applied to an MEC server in a hybrid NOMA-MEC system. The joint resource allocation method for minimizing weighted delay energy includes:

[0007] S1, obtaining the latency requirements of users within the coverage area, and dividing the users into a sensitive user group and a non-sensitive user group according to the latency requirements;

[0008] S2, adding virtual users to the sensitive user group to make the number of members in the sensitive user group the same as that in the non-sensitive user group, and combining the members of the sensitive user group with the members of the non-sensitive user group to form multiple different ST units;

[0009] Among them, each subchannel resource can only be called by one ST unit;

[0010] S3. Create three initial matching lists for recording sensitive users that are not matched, non-sensitive users that are not matched, and sub-channel resources that are not matched respectively;

[0011] S4. Taking the allocated power of each member of each ST unit as a variable, and taking the minimum weighted delay energy of each ST unit under each sub-channel resource as a constraint problem, calculate the weighted delay energy of each ST unit under each sub-channel resource, and sort the weighted delay energy from smallest to largest, and form a preference list of channel resources with the sorted weighted delay energy;

[0012] S5. For any unmatched sub-channel resource in the initial matching list, try to match the ST unit from the best preference to the worst preference according to the order of the weighted delay energy recorded in the preference list. If the unmatched sub-channel resource is successfully matched with the ST unit, delete the corresponding record from the three initial matching lists. If the worst preference fails to match successfully, the matching ends; obtain the matching result of each ST unit and sub-channel resource;

[0013] S6. According to the matching result of the ST unit and the sub-channel resource, obtain the minimum weighted delay energy result of the matching result of the ST unit and the sub-channel resource.

[0014] Advantages of the present invention:

[0015] A joint resource allocation method for minimizing weighted delay energy provided by the present invention comprehensively considers the NOMA user clustering strategy, the inter-cluster frequency band allocation in scheduling, and the in-cluster power allocation control. Group users according to delay requirements, and then form ST units. Taking the allocated power of users as a variable and minimizing the weighted delay energy as a constraint problem, calculate the weighted energy of each ST unit under sub-channel resources. Then, through multiple iterations, match the ST unit with the sub-channel resource. After the matching is completed, each user finds the most suitable resource allocation according to the delay requirement, realizing optimal resource allocation. Compared with the prior art, the present invention has the advantages of saving resources, higher resource allocation efficiency, accurate resource allocation, and better user experience.

[0016] The following will further describe the present invention in detail with reference to the drawings and embodiments. Description of the Drawings

[0017] Figure 1 It is a model diagram of a hybrid NOMA-MEC system used in the present invention;

[0018] Figure 2 It is a flowchart of a joint resource allocation method for minimizing weighted delay energy of the present invention;

[0019] Figure 3It is the process flow chart for implementing step S5 of the present invention;

[0020] Figure 4 It is the curve graph of the system weighted delay energy varying with the delay of sensitive users in the present invention. Specific embodiments

[0021] The following further describes the present invention in detail with reference to specific embodiments, but the implementation manners of the present invention are not limited thereto.

[0022] A joint resource allocation method for minimizing weighted delay energy provided by the present invention is applied to the MEC server in a hybrid NOMA-MEC system. As Figure 1 shown, the system includes 1 MEC server, M sensitive users, N non-sensitive users, and K sub-channel resources. The system has the following characteristics: There is a computing task ready to be offloaded for each user; The MEC server can obtain channel information and user information for use as the basis for policy scheduling; Each sub-channel resource can only be used by one ST unit; The computing time of the computing task in the MEC server and the time for the computing result to be transmitted back are negligible.

[0023] As Figure 2 shown, a joint resource allocation method for minimizing weighted delay energy provided by the present invention includes:

[0024] S1. Obtain the delay requirements of users within the coverage range, and divide the users into a sensitive user group and a non-sensitive user group according to the delay requirements;

[0025] Among them, the sensitive user group and the non-sensitive user group are respectively denoted as U s ={u s1 , u s2 ,..., u sN}, U t ={u t1 , u t2 ,..., u tN}; The number of users in the sensitive group is denoted as M, and the number of users in the non-sensitive group is denoted as N; where M≤N.

[0026] S2. Add virtual users to the sensitive user group to make the number of members in the sensitive user group the same as that in the non-sensitive user group, and combine the members of the sensitive user group and the members of the non-sensitive user group to form multiple different ST units;

[0027] Among them, each sub-channel resource can only be called by one ST unit; N - M virtual users are added to the sensitive user group; any combination of N sensitive users and N non-sensitive users in the network is formed into N×N different ST units. To describe the proposed matching algorithm, a user group composed of 1 sensitive user and 1 non-sensitive user is defined as 1 ST unit. Note that ST units are not independent because they may contain the same sensitive users or non-sensitive users.

[0028] S3. Respectively create three initial matching lists that record unmatched sensitive users, unmatched non-sensitive users, and unmatched sub-channel resources.

[0029] Specifically, S3 includes:

[0030] S31. Take each sub-channel resource as an unmatched sub-channel resource and generate an initial matching list that records unmatched sub-channel resources.

[0031] S32. Take the non-sensitive users corresponding to each ST unit as unmatched non-sensitive users and generate an initial matching list that records unmatched non-sensitive users.

[0032] S33. Take the sensitive users corresponding to each ST unit as unmatched sensitive users and generate an initial matching list that records unmatched sensitive users.

[0033] It should be noted that: the three initial matching lists are SU matchlist , TU matchlist , SC matchlist , to record the flags of whether each sensitive user, non-sensitive user, and sub-channel is matched. By default, all members and sub-channel resources are unmatched at the initial moment.

[0034] S4. Using the allocated power of each member of each ST unit as a variable and taking the minimum weighted delay energy of each ST unit under each sub-channel resource as the objective function, calculate the weighted delay energy of each ST unit under each sub-channel resource, sort the calculated weighted delay energy from small to large, and form a preference list of channel resources with the sorted weighted delay energy.

[0035] Specifically, S4 includes:

[0036] S41. Using the allocated power P si of the sensitive user u si in each ST unit and the allocated power P tj of the non-sensitive user u tj as variables;

[0037] S42. Using the weighted delay energy DET of each ST unit under each sub-channel resourceij is the objective function. Taking the minimization of the objective function as the constrained problem, calculate the weighted delay energy of each ST unit under each sub-channel resource under the constraint conditions, and sort the weighted delay energy from small to large. The sorted weighted delay energy forms a preference list of channel resources.

[0038] When the MEC device receives the NOMA signal, the SIC receiver will perform demodulation. The sensitive user u si and the non-sensitive user u tj The channel gains with the base station on channel k can be expressed as h i,k and g j,k . In the SIC demodulation stage, we choose to demodulate the delay-insensitive user u tj first, and then demodulate the sensitive user u si . Therefore, it is obvious that for the user u tj , there are two stages in its data offloading process. One is the NOMA transmission stage sharing the channel with the user u si . After the calculation task offloading of the user u si is completed, it will enter the OMA transmission stage. Therefore, the user u tj has two data transmission rates in the entire data transmission process, namely the NOMA transmission stage and the OMA transmission stage, which are respectively expressed as:

[0039]

[0040]

[0041] For the user u si , its data transmission rate can be expressed as:

[0042]

[0043] where B k represents the bandwidth of channel k; P si , P tj respectively represent the user u si , u tj in the NOMA stage

[0044] For an ST unit, the total time of its data transmission process can be divided into the NOMA transmission stage and the OMA transmission stage, which are respectively expressed as:

[0045]

[0046] where D si , D tj represent the user u si , u tjThe data volume of task offloading, and the total time can be expressed as:

[0047] τ ij = τ NOMA + τ OMA

[0048] The energy consumed during the whole process can also be expressed in two stages, and the total consumed energy is:

[0049] E ij = (P si + P tj ) × τ NOMA + P tj × τ OMA

[0050] The weighted delay energy can be expressed as:

[0051] DET ij = c1τ ij + c2E ij

[0052] where c = (c1, c2) is the weighting coefficient, and the constraint problem can be expressed as:

[0053] Constraint problem:

[0054]

[0055] Constraint conditions:

[0056] C1: τ NOMA ≤ T si

[0057] C2: τ NOMA + τ OMA ≤ T tj

[0058] C3: 0 ≤ P si ≤ P max

[0059] C4: 0 ≤ P tj ≤ P max

[0060] The objective function is as follows:

[0061]

[0062] δ is the noise standard deviation; α is an artificial parameter introduced to simplify the expression, and its value can be expressed as:

[0063]

[0064] It can be seen that the objective function is non - convex and the optimization variables are highly coupled. To simplify the optimization process, the following transformation is performed on this optimization problem. According to the known conditions, it can be obtained that:

[0065]

[0066] Among them:

[0067]

[0068] After the transformation, the objective function can be simplified to:

[0069]

[0070] Among them:

[0071]

[0072] S5. For any unmatched sub - channel resource in the initial matching list, try to match the ST unit from the best preference to the worst preference according to the order of the weighted delay energy recorded in the preference list. If the unmatched sub - channel resource is successfully matched with the ST unit, delete the corresponding record from the three initial matching lists. If the worst preference fails to match successfully, the matching ends; obtain the matching results of each ST unit and sub - channel resource;

[0073] Specifically, referring to Figure 3 , S5 includes:

[0074] S51. In each iteration process, for any unmatched sub - channel resource k in the initial matching list, according to the ascending order recorded in the preference list, and send a matching request from the ST units that have been rejected in the previous iteration process;

[0075] S52. If the member of the ST unit receiving the matching request has never received any offer before the current iteration, temporarily match any unmatched sub - channel resource k with this ST unit;

[0076] S53. If the member u of the ST unit si or u tj has at least one that has been matched with the sub - channel resource before the current iteration, further determine whether the unmatched sub - channel resource and this ST unit are the optimal match;

[0077] S54. If the unmatched sub - channel resource and this ST unit are the optimal match in S53, then this sub - channel resource k preempts this ST unit and reaches a match with it so that this ST unit rejects the previous match, obtaining an α - type blocking triple formed by this ST unit and the sub - channel resource;

[0078] S55. If the unmatched sub-channel resource in S53 is not the optimal match for the ST unit, send a matching request to the next ST unit of the ST unit, and execute S52 to S55;

[0079] S56. Remove the ST unit and sub-channel resource that have achieved a match from the initial match table, and re-record the preempted ST unit in the initial match table;

[0080] S57. Repeat S51 to S56 in each iteration until all ST units form a match, or until the last ST unit, end the iteration, and obtain the matching result of the ST unit and the sub-channel resource.

[0081] S6. According to the matching result of the ST unit and the sub-channel resource, obtain the minimum weighted delay energy result of the matching result of the ST unit and the sub-channel resource.

[0082] Reference Figure 3 , starting from the best preference, that is, starting from the ST with the minimum weighted delay energy until the last ST ends, determine whether the first ST unit is occupied. If not, determine whether the unmatched sub-channel resource forms an optimal match with the ST unit, and then delete the user and sub-channel resource corresponding to the optimal match in the three initial match lists; if any user is occupied in the first ST unit, further determine whether the unmatched sub-channel resource is the optimal match for the ST unit. If so, preempt the ST unit and form an α-type blocking triple with the ST unit, and re-put the preempted sub-channel resource into the initial match list; if the unmatched sub-channel resource is not the optimal match for the ST unit, abandon the ST unit, apply to access the next ST unit to perform the optimal match until the last ST unit, and obtain the matching result of the ST unit and the sub-channel resource.

[0083] In addition, each ST unit that temporarily reserves the offer of sub-channel k is not yet ready to actually accept the offer (match) because if other sub-channels also make proposals to it, it still has the right to choose a better offer in the next iteration. When no sub-channel sends an application to the ST unit, this means that all sub-channels either match the ST unit or are rejected by all ST units, and the iteration stops.

[0084] Reference Figure 4 , Figure 4 is the curve graph of the system weighted delay energy of the present invention changing with the delay of sensitive users. It can be seen from Figure 4 that in the traditional scheme, as the delay of sensitive users increases, the weighted delay energy gradually increases. In this application, as the delay of sensitive users increases, the weighted delay energy decreases significantly. Therefore, the present invention has better resource allocation and higher user experience compared with the prior art.

[0085] A joint resource allocation method for minimizing weighted delay energy provided by the present invention comprehensively considers the NOMA user clustering strategy, the inter-cluster frequency band allocation in scheduling, and the intra-cluster power allocation control. The users are grouped according to the delay requirements, and then ST units are formed. Taking the allocated power of the users as variables and minimizing the weighted delay energy as a constraint problem, the weighted energy of each ST unit under the sub-channel resources is obtained. Then, through multiple iterations, the ST units are matched with the sub-channel resources. After the matching is completed, each user finds the most suitable resource allocation according to the delay requirements, achieving optimal resource allocation. Compared with the prior art, the present invention has the advantages of resource saving, high resource allocation efficiency, accurate resource allocation, and better user experience.

[0086] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality" means two or more unless otherwise specifically defined.

[0087] Although the present application has been described in connection with various embodiments herein, however, in implementing the claimed present application, those skilled in the art can understand and realize other variations of the disclosed embodiments by viewing the accompanying drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "one" does not exclude a plurality of cases.

[0088] The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited only to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A joint resource allocation method for minimizing weighted delay energy, characterized in that The MEC server applied to the system of hybrid NOMA-MEC, and the joint resource allocation method for minimizing weighted delay energy includes: S1. Obtain the delay requirements of users within the coverage range, and divide the users into a sensitive user group and a non-sensitive user group according to the delay requirements; S2. Add virtual users to the sensitive user group to make the number of members in the sensitive user group the same as that in the non-sensitive user group, and combine the members of the sensitive user group and the members of the non-sensitive user group to form multiple different ST units; Among them, each sub-channel resource can only be called by one ST unit; S3. Create three initial matching lists that record the unmatched sensitive users, unmatched non-sensitive users, and unmatched sub-channel resources respectively; S4. Take the allocated power of each member of each ST unit as a variable, and take the minimum weighted delay energy of each ST unit under each sub-channel resource as a constraint problem, calculate the weighted delay energy of each ST unit under each sub-channel resource, sort the weighted delay energy from small to large, and form a preference list of channel resources with the sorted weighted delay energy; S5. For any unmatched sub-channel resource in the initial matching list, try to match the ST unit from the best preference to the worst preference according to the order of the weighted delay energy recorded in the preference list. If the unmatched sub-channel resource successfully matches the ST unit, delete the corresponding records from the three initial matching lists. If the worst preference fails to match successfully, the matching ends; obtain the matching results of each ST unit and sub-channel resource; S6. According to the matching results of the ST unit and the sub-channel resource, obtain the minimum weighted delay energy result of the matching results of the ST unit and the sub-channel resource; S4 includes: S41, with the allocated power P of the sensitive user u in each ST unit si and the allocated power P of the non-sensitive user u si as variables; tj tj ​​ S42, using the weighted delay energy DET of each ST unit under each sub-channel resource ij as the objective function, taking the minimization of the objective function as a constrained problem, calculating the weighted delay energy of each ST unit under each sub-channel resource under the constraint conditions, sorting the weighted delay energies from smallest to largest, and forming a preference list of channel resources with the sorted weighted delay energies; In S42, the objective function is: The constraint problem is: The constraint conditions are: C1: τ NOMA ≤ T si C2: τ NOMA + τ OMA ≤ T tj C3: 0 ≤ P si ≤ P max C4: 0 ≤ P tj ≤ P max User u tj During the entire data transmission process, there are two data transmission rates in the NOMA transmission stage and the OMA transmission stage, which are respectively expressed as: User u si The transmission rate of the entire data of can be expressed as: Among them, B k represents the bandwidth of channel k; P si , P tj respectively represent the transmission powers of users u si , u tj in the NOMA stage; P max represents the maximum power, that is, the transmission power of user u tj in the OMA stage; For an ST unit, the total time of its data transmission process is divided into a NOMA transmission stage and an OMA transmission stage, which are respectively expressed as: Among them, D si and D tj respectively represent the amount of data offloaded for tasks of users u si and u tj ; τ ij =τ NOMA +τ OMA represents the total transmission time of the ST unit; The total energy consumed in the whole process is: E ij = (P si + P tj ) × τ NOMA + P tj × τ OMA Among them, the sensitive user u si and the non-sensitive user u tj The channel gains with the base station on channel k can be expressed as h i,k and g j,k ; c = (c1, c2) is the weighting coefficient, D si , D tj represents the time taken for the complete process of task offloading of user u si , u tj ; δ is the noise standard deviation; α is an artificial parameter introduced to simplify the expression, and its value can be expressed as:

2. The joint resource allocation method for minimizing weighted delay energy according to claim 1, wherein In S1, the sensitive user group and the non-sensitive user group are respectively denoted as U s ={u s1 , u s2 ,..., u sM}, U t ={u t1 , u t2 ,..., u tN}; the number of users in the sensitive group is denoted as M, and the number of users in the non-sensitive group is denoted as N; where M ≤ N, S2 includes: adding N - M virtual users to the sensitive user group; combining any of the N sensitive users and N non-sensitive users in the network to form N×N different ST units.

3. A joint resource allocation method for minimizing weighted delay energy according to claim 1, characterized in that S3 includes: Regarding each sub-channel resource as an unmatched sub-channel resource, generate an initial matching list that records the unmatched sub-channel resources; Regarding the non-sensitive user corresponding to each ST unit as an unmatched non-sensitive user, generate an initial matching list that records the unmatched non-sensitive users; Regarding the sensitive user corresponding to each ST unit as an unmatched sensitive user, generate an initial matching list that records the unmatched sensitive users.

4. The joint resource allocation method for minimizing weighted delay energy according to claim 1, wherein S5 Includes: S51. In each iteration process, for any unmatched sub-channel resource k in the initial matching list, according to the order from small to large recorded in the preference list, and send a matching request from the ST units that have been rejected in the previous iteration process; S52. If the members of the ST unit receiving the matching request have never received any offers before the current iteration, temporarily match any unmatched sub-channel resource k with this ST unit; S53, if member u of the ST unit si or u tj has had at least one match with the sub-channel resource before the current iteration, then further determine whether the unmatched sub-channel resource is the optimal match with this ST unit; S54. If the unmatched sub-channel resource in S53 is not the optimal match with the ST unit, then this sub-channel resource k preempts this ST unit and reaches a match with it so that this ST unit rejects the previous match, obtaining an α-type blocking triple formed by this ST unit and the sub-channel resource; S55. If the unmatched sub-channel resource in S53 is not the optimal match with the ST unit, then send a match request to the next ST unit of this ST unit, and execute S52 to S55; S56. Remove the ST unit that has reached a match and the sub-channel resource from the initial match table, and re-record the preempted ST unit in the initial match table; S57. Repeat S51 to S56 in each iteration until all ST units form matches, or until the last ST unit, ending the iteration and obtaining the matching result of the ST units and the sub-channel resources.

Images