TWI732350B - Resource allocation method and data control center based on genetic algorithm - Google Patents
Resource allocation method and data control center based on genetic algorithm Download PDFInfo
- Publication number
- TWI732350B TWI732350B TW108142115A TW108142115A TWI732350B TW I732350 B TWI732350 B TW I732350B TW 108142115 A TW108142115 A TW 108142115A TW 108142115 A TW108142115 A TW 108142115A TW I732350 B TWI732350 B TW I732350B
- Authority
- TW
- Taiwan
- Prior art keywords
- chromosome
- chromosomes
- candidate
- frequency band
- energy
- Prior art date
Links
Images
Abstract
Description
本發明是有關於一種通訊資源分配機制,且特別是有關於一種基於基因演算法的資源分配方法及資料控制中心。 The present invention relates to a communication resource allocation mechanism, and in particular to a resource allocation method and data control center based on genetic algorithm.
請參照圖1,其是習知的通訊系統架構示意圖。如圖1所示,通訊系統100包括資料控制中心110、主要(primary)用戶裝置120、主要基地台125、次要(secondary)用戶裝置130及次要基地台135。在圖1中,主要基地台125(例如eNB)可允許主要用戶裝置120(例如一般的手機)基於經授權頻譜(licensed spectrum)而實現通訊功能,而次要基地台125可用於讓次要用戶裝置130基於未經授權頻譜(unlicensed spectrum)而實現通訊功能。或者,次要用戶裝置130也可自行基於未經授權頻譜進行例如藍牙或是其他類似的通訊行為。
Please refer to FIG. 1, which is a schematic diagram of a conventional communication system architecture. As shown in FIG. 1, the
隨著近年物聯網(Internet of Things,IoT)相關產業興起,IoT電子設備之無線傳輸需求日益增加,也相應導致IoT的能 源及頻譜資源越趨缺乏。為了使上述資源的運用更具彈性,現有技術中已有基於感知無線電(cognitive radio,CR)的IoT技術。 With the rise of Internet of Things (IoT)-related industries in recent years, the demand for wireless transmission of IoT electronic devices has increased, which has also led to the power of IoT. Sources and spectrum resources are increasingly scarce. In order to make the use of the above resources more flexible, there is an IoT technology based on cognitive radio (CR) in the prior art.
在圖1情境中,資料控制中心110可基於頻譜租借(spectrum leasing)的概念而從主要用戶裝置120所提供的資料而得知當下有哪些空閒的子通道(或稱頻段),進而可據以透過一定的資源分配手段將這些資源分配予基於CR的IoT裝置(例如圖1中位於虛線圈中的裝置)進行通訊。
In the scenario of FIG. 1, the
在相關的現有資源分配手段中,大致可區分為以下數類作法:(1)透過迭代計算尋找平衡點以求解;(2)將問題轉化為凸最佳化問題(convex optimization problem),並用相關演算法求解。然而,上述作法屬於單一目標最佳化,故無法解決多個目標函數可能彼此矛盾的情形。 Related existing resource allocation methods can be roughly divided into the following types of practices: (1) Find a balance point to solve through iterative calculation; (2) Convert the problem into a convex optimization problem, and use the correlation Algorithm solution. However, the above method is a single objective optimization, so it cannot solve the situation where multiple objective functions may contradict each other.
另一方面,相關現有技術中亦有討論多目標最佳化的文獻,其作法可區分為以下數類:(1)將多目標最佳化問題轉換為單一目標最佳化問題求解,例如ε限制(ε-constraint)方法。然而,此種作法所獲得之最佳解效能容易受ε值所影響,故其效能並不穩定;(2)利用多目標進化演算法求解,例如SPEA-II以及NSGA-II方法。然而,此類做法在運行過程中無法確保產生可行解,導致其收斂效能不佳。 On the other hand, there are also documents discussing multi-objective optimization in the related prior art. The methods can be divided into the following categories: (1) Convert the multi-objective optimization problem into a single-objective optimization problem, such as ε Constraint (ε-constraint) method. However, the best solution performance obtained by this method is easily affected by the value of ε, so its performance is not stable; (2) Solving by multi-objective evolutionary algorithms, such as SPEA-II and NSGA-II methods. However, such an approach cannot ensure that a feasible solution is produced during operation, resulting in poor convergence performance.
此外,現有技術的相關資源分配方法僅考慮靜態通訊環境,故並不適用於實際的動態通訊環境。具體而言,現有技術係假設相關的系統參數(例如可用通道數量等)不隨時間變化,但 這些系統參數實際上應會隨著時間而改變。換言之,只要系統參數發生變化,現有技術即必須重新運算以求得當下的最佳解,導致收斂效率不佳。 In addition, the related resource allocation method in the prior art only considers the static communication environment, so it is not suitable for the actual dynamic communication environment. Specifically, the prior art assumes that relevant system parameters (such as the number of available channels, etc.) do not change over time, but These system parameters should actually change over time. In other words, as long as the system parameters change, the prior art must recalculate to find the best solution at the moment, resulting in poor convergence efficiency.
有鑑於此,本發明提供一種基於基因演算法的資源分配方法及資料控制中心,其可用於解決上述技術問題。 In view of this, the present invention provides a resource allocation method and a data control center based on genetic algorithm, which can be used to solve the above technical problems.
本發明提供一種基於基因演算法的資源分配方法,適於管理多個次要用戶裝置的一資料控制中心。所述方法包括:在第t個時間點判斷所述多個次要用戶裝置所處的一通訊環境是否改變,其中所述多個次要用戶裝置之間存在多個傳輸鏈結,且t為大於1的正整數;反應於判定所述第t個時間點的通訊環境已改變,依據至少一歷史候選染色體集合產生對應於第t個時間點的一候選染色體集合,其中候選染色體集合包括多個候選染色體,其中各候選染色體包括各傳輸鏈結對應的一候選能源及一候選頻段;對候選染色體集合執行一多目標最佳化演算法,以對各傳輸鏈結分配一最佳能源及一最佳頻段;控制所述多個次要用戶裝置依據各傳輸鏈結對應的最佳能源及最佳頻段進行通訊。 The present invention provides a resource allocation method based on genetic algorithm, which is suitable for managing a data control center of multiple secondary user devices. The method includes: judging at the t-th time point whether a communication environment in which the multiple secondary user devices are located has changed, wherein there are multiple transmission links between the multiple secondary user devices, and t is A positive integer greater than 1; in response to determining that the communication environment at the t-th time point has changed, a candidate chromosome set corresponding to the t-th time point is generated based on at least one historical candidate chromosome set, wherein the candidate chromosome set includes a plurality of Candidate chromosomes, where each candidate chromosome includes a candidate energy and a candidate frequency band corresponding to each transmission link; a multi-objective optimization algorithm is performed on the set of candidate chromosomes to allocate an optimal energy and a maximum to each transmission link Optimal frequency band; controlling the multiple secondary user devices to communicate according to the optimal energy and optimal frequency band corresponding to each transmission link.
本發明提供一種基於基因演算法分配資源的資料控制中心,用於管理多個次要用戶裝置。所述資料控制中心包括收發器及處理器。收發器從多個主要用戶裝置接收指示多個空閒頻段的的多個資料。處理器耦接收發器並經配置以:基於所述多個資料 估計所述多個主要用戶裝置與所述多個次要用戶裝置所處的一通訊環境;在第t個時間點判斷所述多個次要用戶裝置所處的通訊環境是否改變,其中所述多個次要用戶裝置之間存在多個傳輸鏈結,且t為大於1的正整數;反應於判定所述第t個時間點的通訊環境已改變,依據至少一歷史候選染色體集合產生對應於第t個時間點的一候選染色體集合,其中候選染色體集合包括多個候選染色體,其中各候選染色體包括各傳輸鏈結對應的一候選能源及一候選頻段;對候選染色體集合執行一多目標最佳化演算法,以對各傳輸鏈結分配一最佳能源及一最佳頻段;控制所述多個次要用戶裝置依據各傳輸鏈結對應的最佳能源及最佳頻段進行通訊。 The invention provides a data control center that allocates resources based on genetic algorithm, which is used to manage multiple secondary user devices. The data control center includes a transceiver and a processor. The transceiver receives a plurality of data indicating a plurality of idle frequency bands from a plurality of main user devices. The processor is coupled to the receiver and is configured to: based on the plurality of data Estimate a communication environment in which the plurality of primary user devices and the plurality of secondary user devices are located; determine at the t-th time point whether the communication environment in which the plurality of secondary user devices are located has changed, wherein There are multiple transmission links between multiple secondary user devices, and t is a positive integer greater than 1; in response to determining that the communication environment at the t-th time point has changed, a set of at least one historical candidate chromosome corresponding to A candidate chromosome set at the t-th time point, where the candidate chromosome set includes multiple candidate chromosomes, and each candidate chromosome includes a candidate energy source and a candidate frequency band corresponding to each transmission link; perform a multi-objective optimization on the candidate chromosome set The optimization algorithm is used to allocate an optimal energy source and an optimal frequency band to each transmission link; and control the multiple secondary user devices to communicate according to the optimal energy and optimal frequency band corresponding to each transmission link.
為讓本發明的上述特徵和優點能更明顯易懂,下文特舉實施例,並配合所附圖式作詳細說明如下。 In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail in conjunction with the accompanying drawings.
100:通訊系統 100: Communication system
110、210:資料控制中心 110, 210: Data Control Center
120、220:主要用戶裝置 120, 220: main user device
125:主要基地台 125: main base station
130、230:次要用戶裝置 130, 230: secondary user device
135:次要基地台 135: Secondary Base Station
1~10:節點 1~10: Node
212:收發器 212: Transceiver
214:處理器 214: processor
510、520:成員染色體 510, 520: member chromosomes
610、610a、620、620a:參考染色體 610, 610a, 620, 620a: reference chromosome
S710~S740:步驟 S710~S740: steps
P (i,j):能源 P ( i,j ) : energy
c (i,j):頻段 c ( i,j ) : frequency band
Γ(t,k)| t=1:染色體集合 Γ ( t,k )| t =1 : chromosome set
Γ'(t,k)| t=1:第一集合 Γ' ( t,k )| t =1 : the first set
Γ"(t,k)| t=1:第二集合 Γ" ( t,k )| t =1 : the second set
Γ'''(t,k)| t=1、Γ'''(t,k max )| t=1:第三集合 Γ''' ( t,k )| t =1 , Γ''' ( t,k max )| t =1 : the third set
x 1(t)| t=1~x 50(t)| t=1、x 1 '(t)| t=1~x 50 '(t)| t=1、x 1 "(t)| t=1~x 50 "(t)| t=1、x 1 '''(t)| t=1~x 50 '''(t)| t=1、x *(t)| t=1:染色體 x 1 ( t )| t =1 ~ x 50 ( t )| t =1 , x 1 ' ( t )| t =1 ~ x 50 ' ( t )| t =1 , x 1 " ( t )| t =1 ~ x 50 " ( t )| t =1 , x 1 ''' ( t )| t =1 ~ x 50 ''' ( t )| t =1 , x * ( t )| t =1 : chromosome
M(t):可用頻譜集合 M( t ): Available spectrum set
L:傳輸鏈結集合 L : Transmission link collection
F:資料流集合 F : Data stream collection
圖1是習知的通訊系統架構示意圖。 Figure 1 is a schematic diagram of a conventional communication system architecture.
圖2是依據本發明之一實施例繪示的系統示意圖。 Fig. 2 is a schematic diagram of a system according to an embodiment of the present invention.
圖3是依據本發明之一實施例繪示的次要用戶裝置之間的能源及頻譜分配示意圖。 FIG. 3 is a schematic diagram of energy and spectrum allocation among secondary user devices according to an embodiment of the present invention.
圖4是依據本發明第一實施例繪示的多目標最佳化演算法應用情境示意圖。 FIG. 4 is a schematic diagram of an application scenario of a multi-object optimization algorithm according to the first embodiment of the present invention.
圖5是依據本發明第一實施例繪示的產生混洗版本的示意 圖。 Fig. 5 is a schematic diagram of generating a shuffled version according to the first embodiment of the present invention Figure.
圖6是依據本發明第一實施例繪示的調整第一參考染色體的示意圖。 Fig. 6 is a schematic diagram of adjusting the first reference chromosome according to the first embodiment of the present invention.
圖7是依據本發明第二實施例繪示的基於基因演算法的資源分配方法流程圖。 Fig. 7 is a flowchart of a resource allocation method based on a genetic algorithm according to a second embodiment of the present invention.
圖8是依據本發明之一實施例繪示的資料控制中心功能方塊圖。 Fig. 8 is a functional block diagram of a data control center according to an embodiment of the present invention.
請參照圖2,其是依據本發明之一實施例繪示的系統示意圖。概略而言,在本實施例中,在第t個時間點(t為大於等於1的正整數)時,主要用戶裝置220可向資料控制中心210提供所測得的可用頻譜集合M(t)。在一實施例中,此可用頻譜集合M(t)例如可包括一或多個空閒的子通道或頻段,而其總數可代稱為可用通道數量,但可不限於此。次要用戶裝置230(例如是CR裝置)可用以向資料控制中心210提供傳輸鏈結集合L及資料流集合F。基於可用頻譜集合M(t)、傳輸鏈結集合L及資料流集合F,資料控制中心210可執行本發明提出的基於基因演算法的資源分配方法,以對次要用戶裝置230進行資源分配。在本發明的實施例中,資料控制中心210例如可用於對次要用戶裝置230分配能源及頻段等資源,藉以讓次要用戶裝置230可使用所分配的能源而在所分配的頻段上進行傳輸。
Please refer to FIG. 2, which is a schematic diagram of a system according to an embodiment of the present invention. Roughly speaking, in this embodiment, at the t-th time point (t is a positive integer greater than or equal to 1), the
請參照圖3,其是依據本發明之一實施例繪示的次要用戶裝置之間的能源及頻譜分配示意圖。在圖3所示情境中,假設共有10個次要用戶裝置,其個別對應於節點1~10,並可在經本發明的資料控制中心210分配能源及頻段之後,據以進行通訊。
Please refer to FIG. 3, which is a schematic diagram of energy and spectrum allocation among secondary user devices according to an embodiment of the present invention. In the scenario shown in FIG. 3, it is assumed that there are 10 secondary user devices, which respectively correspond to nodes 1-10, and can communicate with each other after energy and frequency bands are allocated by the
在本實施例中,P (i,j)代表由節點i傳輸信號至節點j所使用的能源,而c (i,j)則代表由節點i傳輸信號至節點j所使用的頻段。此外,f代表傳輸資料流,而由圖3可看出,其中存在傳輸資料流f 1~f 3。基此,節點1~10可將以上資訊作為圖1中的傳輸鏈結集合L及資料流集合F而提供至資料控制中心210,但本發明可不限於此。另外,為便於說明,以下將節點i及節點j之間的傳輸鏈結以L (i,j)代稱。
In this embodiment, P ( i, j ) represents the energy used by node i to transmit signals to node j, and c ( i , j ) represents the frequency band used by node i to transmit signals to node j. In addition, f represents a transmission data stream, and it can be seen from Figure 3 that there are transmission data streams f 1 to f 3 . Based on this, the nodes 1-10 can provide the above information as the transmission link set L and the data stream set F in FIG. 1 to the
概略而言,在所考慮的每個時間點中,本發明的方法可先經由一定的初始化操作產生一個染色體集合,其可包括多個染色體,而各染色體可包括對應於各個傳輸鏈結L (i,j)的P (i,j)及c (i,j)。之後,本發明的方法可再對此染色體集合進行本發明提出的多目標最佳化演算法,以求得各傳輸鏈結L (i,j)所對應的最佳資源分配解,亦即最佳能源及最佳頻段,藉以讓相應的節點(即,次要用戶裝置)可據以進行傳輸。 Generally speaking, at each time point under consideration, the method of the present invention may first generate a chromosome set through a certain initialization operation, which may include multiple chromosomes, and each chromosome may include a link corresponding to each transmission link L ( i, j) of P (i, j) and c (i, j). After that, the method of the present invention can perform the multi-objective optimization algorithm proposed by the present invention on this chromosome set to obtain the optimal resource allocation solution corresponding to each transmission link L ( i, j ) , that is, the most The best energy source and the best frequency band, so that the corresponding node (ie, the secondary user device) can transmit accordingly.
在本發明的實施例中,由於各時間點所對應的歷史資料量及通訊環境可能不盡相同,故各時間點相關的初始化操作亦有所不同,使得所產生的染色體集合的態樣亦有所不同。儘管如此,在各時間點所使用的皆為同一個多目標最佳化演算法,而多目標 最佳化演算法的細節將在之後另行說明。 In the embodiment of the present invention, since the amount of historical data and communication environment corresponding to each time point may be different, the initialization operations related to each time point are also different, so that the generated chromosome set also has different patterns. The difference. Nevertheless, the same multi-objective optimization algorithm is used at each time point, and the multi-objective optimization algorithm The details of the optimization algorithm will be explained later.
在第一實施例中,在第1個時間點(可理解為t等於1)時,由於尚未有任何歷史資料,亦無從判斷次要用戶裝置(例如圖2中的節點1~10)所處的通訊環境是否改變,故資料控制中心210可隨機產生初始染色體集合,其中初始染色體集合包括隨機產生的多個初始染色體,且各初始染色體包括各傳輸鏈結L (i,j)對應的初始能源及初始頻段。之後,資料控制中心210可對初始染色體集合執行多目標最佳化演算法,以對各傳輸鏈結L (i,j)分配初始最佳能源及初始最佳頻段。接著,資料控制中心210可控制次要用戶裝置依據各傳輸鏈結L (i,j)對應的初始最佳能源及初始最佳頻段進行通訊。
In the first embodiment, at the first time point (it can be understood as t equals 1), since there is no historical data, it is impossible to determine where the secondary user device (such as
為便於說明在第1個時間點時對初始染色體集合所執行的多目標最佳化演算法的細節,以下另輔以圖4進行說明。 In order to facilitate the description of the details of the multi-objective optimization algorithm performed on the initial chromosome set at the first time point, the following description is supplemented with FIG. 4.
請參照圖4,其是依據本發明第一實施例繪示的多目標最佳化演算法應用情境示意圖。在本實施例中,多目標最佳化演算法可包括k max 個迭代操作(k max 為一正整數)。 Please refer to FIG. 4, which is a schematic diagram of the application scenario of the multi-objective optimization algorithm according to the first embodiment of the present invention. In the present embodiment, the multi-objective optimization algorithm may include two iterations k max (k max is a positive integer).
首先,由圖4可看出,隨機產生的初始染色體集合Γ(t,k)| t=1可包括多個初始染色體x 1(t)| t=1~x 50(t)| t=1,且各初始染色體x 1(t)| t=1~x 50(t)| t=1可包括對應於各傳輸鏈結L (i,j)的初始能源及初始頻段。 First, it can be seen from Figure 4 that the randomly generated initial chromosome set Γ ( t,k )| t =1 can include multiple initial chromosomes x 1 ( t )| t =1 ~ x 50 ( t )| t =1 , And each initial chromosome x 1 ( t )| t =1 ~ x 50 ( t )| t =1 may include the initial energy and initial frequency band corresponding to each transmission link L ( i,j ).
以初始染色體x 1(t)| t=1為例,其可包括對應於傳輸鏈結L (1,2)至L (9,10)的初始頻段c 1,(1,2)( t )~ c 1,(9,10)( t )及初始能源P 1,(1,2)( t )至 P 1,(9,10) (t),而初始頻段c 1,(1,2) (t)~ c 1,(9,10) (t)及初始能源P 1,(1,2) (t)至 P 1,(9,10) (t)個別皆為隨機產生。在一實施例中,各初始頻段例如是介於0及1之間的數值,而各初始能源例如是介於能源下限值(以P min 表示)及能源上限值(以P max 表示)之間的數值。基此,本領域具通常知識者應可理解其他初始染色體的結構所代表的意義,於此不另贅述。 Taking the initial chromosome x 1 ( t )| t =1 as an example, it may include the initial frequency band c 1,(1,2) ( t ) corresponding to the transmission link L (1,2) to L (9,10) ~ c 1 ,( 9 , 10 ) ( t ) and the initial energy P 1,(1,2) ( t ) to P 1,(9,10) ( t ) , and the initial frequency band c 1,(1,2) ( t ) ~ c 1,(9,10) ( t ) and the initial energy P 1,(1,2) ( t ) to P 1,(9,10) ( t ) are all randomly generated. In one embodiment, each initial frequency band is, for example, a value between 0 and 1, and each initial energy is, for example, between the lower limit of energy ( represented by P min ) and the upper limit of energy ( represented by P max ). The value between. Based on this, those with ordinary knowledge in the field should be able to understand the meaning represented by the structure of other initial chromosomes, and will not be repeated here.
另外,在本發明實施例中,設計者可依需求而將初始染色體集合中的初始染色體數量設定為所需的預設數量。以圖4為例,其係假設初始染色體集合中共包括50個(即,預設數量)初始染色體。並且,為便於理解,本發明係將所有提到的集合中的染色體的預設數量皆假設為50,但其並非用以限定本發明可能的實施方式。 In addition, in the embodiment of the present invention, the designer can set the initial number of chromosomes in the initial chromosome set to the required preset number according to requirements. Taking Fig. 4 as an example, it is assumed that the initial chromosome set includes a total of 50 (ie, the preset number) initial chromosomes. In addition, for ease of understanding, the present invention assumes that the preset number of chromosomes in all mentioned sets is 50, but it is not used to limit the possible embodiments of the present invention.
在本實施例中,在多目標最佳化演算法的第k個(1k k max )迭代操作中,資料控制中心210可對特定染色體集合採用比較選取法(tournament selection),以產生第一集合Γ'(t,k)| t=1。在不同的實施例中,若k為1,則上述特定染色體集合為初始染色體集合Γ(t,k)| t=1,而若k不為1,則上述特定染色體集合可為第k-1個迭代操作所產生的結果,之後將作進一步說明。
In this embodiment, in the kth (1 k In the iterative operation of k max ), the
為便於理解,以下將先針對第1個(即,k為1)迭代操作的細節進行說明。如圖4所示,第一集合Γ'(t,k)| t=1包括多個第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1。並且,各第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1可包括對應於各傳輸鏈結L (i,j)的第一能源及第 一頻段。 For ease of understanding, the details of the first (that is, k is 1) iterative operation will be described below. As shown in Figure 4, the first set Γ' ( t,k )| t =1 includes multiple first chromosomes x 1 ' ( t )| t =1 ~ x 50 ' ( t )| t =1 . Moreover, each first chromosome x 1 ′ ( t )| t =1 ~ x 50 ′ ( t )| t =1 may include the first energy source and the first frequency band corresponding to each transmission link L ( i,j ).
以第一染色體x 1 '(t)| t=1為例,其可包括對應於傳輸鏈結L (1,2)至L (9,10)的第一頻段c 1,(1,2) ' (t)~ c 1,(9,10) ' (t)及第一能源P 1,(1,2) ' (t)至 P 1,(9,10) '(t)。基此,本領域具通常知識者應可理解其他第一染色體的結構所代表的意義,於此不另贅述。 Taking the first chromosome x 1 ' ( t )| t =1 as an example, it may include the first frequency band c 1,(1,2) corresponding to the transmission link L (1,2) to L (9,10) ' ( t ) ~ c 1,(9,10) ' ( t ) and the first energy P 1,(1,2) ' ( t ) to P 1,(9,10) ' ( t ) . Based on this, a person with ordinary knowledge in the field should understand the meaning represented by the structure of other first chromosomes, which will not be repeated here.
如上所述,各第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1係基於比較選取法而產生,以下針對比較選取法作進一步說明。在一實施例中,在取得初始染色體集合Γ(t,k)| t=1作為上述特定染色體集合之後,資料控制中心210可依序執行以下步驟:(a)將第一集合Γ'(t,k)| t=1初始化為空集合;(b)隨機取出初始染色體集合Γ(t,k)| t=1中的初始染色體x 1(t)| t=1~x 50(t)| t=1的其中之二者作為第一比較對象及第二比較對象;(c)將第一比較對象及第二比較對象的其中之一者加入第一集合Γ'(t,k)| t=1,以作為第一染色體之一;(d)重複步驟(b)及(c),直至第一集合Γ'(t,k)| t=1中的第一染色體的數量達到預設數量(例如50)。
As mentioned above, each first chromosome x 1 ' ( t )| t =1 ~ x 50 ' ( t )| t =1 is generated based on the comparative selection method. The following will further explain the comparative selection method. In one embodiment, after obtaining the initial chromosome set Γ ( t, k )| t =1 as the above-mentioned specific chromosome set, the
在一實施例中,在以上的步驟(c)中,資料控制中心210可決定第一比較對象的第一擁擠距離及第二比較對象的第二擁擠距離,其中第一擁擠距離關聯於第一比較對象的能源使用率、通道使用公平性及頻譜使用率,而第二擁擠距離關聯於第二比較對象的能源使用率、通道使用公平性及頻譜使用率。之後,反應於判定第一擁擠距離大於第二擁擠距離,資料控制中心210可將第一比較對象加入第一集合Γ'(t,k)| t=1,反之則可將第二比較對象加
入第一集合Γ'(t,k)| t=1。
In one embodiment, in the above step (c), the
在一實施例中,假設第一比較對象可廣泛地表徵為x(t),則第一比較對象的第一擁擠距離可表徵為:
在一實施例中,F EE (x(t))即為x(t)對應的能源使用率,F Fair (x(t))為x(t)對應的通道使用公平性,F Spec (x(t))為x(t)對應的頻譜使用率。並且,F EE (x(t))、F Fair (x(t))及F Spec (x(t))可分別計算如下:
相似地,第二比較對象的第二擁擠距離亦可基於以上教示而求得,於此不另贅述。另外,若第一(或第二)比較對象為初始染色體x 1(t)| t=1或x 50(t)| t=1(即,最首或最末的初始染色體),則其相應的第一(或第二)擁擠距離可定義為無限大,但本發明可不限於此。 Similarly, the second congestion distance of the second comparison object can also be obtained based on the above teachings, which will not be repeated here. In addition, if the first (or second) comparison object is the initial chromosome x 1 ( t )| t =1 or x 50 ( t )| t =1 (that is, the first or last initial chromosome), then the corresponding The first (or second) congestion distance of can be defined as infinite, but the present invention may not be limited to this.
在依據以上教示取得第一集合Γ'(t,k)| t=1中的第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1之後,資料控制中心210可對第一集合Γ'(t,k)| t=1重複執行交叉(crossover)機制或變異(mutation)機制,以產生第二集合Γ"(t,k)| t=1,其中第二集合Γ"(t,k)| t=1可包括多個第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1,且各第二染色體x 1 "(t)| t=1~x 50"(t)| t=1包括對應於各傳輸鏈結L (i,j)的第二能源及第二頻段。
After obtaining the first chromosome x 1 ' ( t )| t =1 ~ x 50 ' ( t )| t =1 in the first set Γ' ( t,k )| t =1 according to the above teaching, the
以第二染色體x 1 "(t)| t=1為例,其可包括對應於傳輸鏈結L (1,2)至L (9,10)的第二頻段c 1,(1,2) "( t )~ c 1,(9,10) "( t )及第二能源P 1,(1,2) "( t )至 P 1,(9,10) "(t)。基此,本領域具通常知識者應可理解其他第二染色體的結構所代表的意義,於此不另贅述。 Taking the second chromosome x 1 " ( t )| t =1 as an example, it may include the second frequency band c 1,(1,2) corresponding to the transmission link L (1,2) to L (9,10) " ( t )~ c 1,(9,10) " ( t ) and the second energy P 1,(1,2) " ( t ) to P 1,(9,10) " ( t ) . Based on this, Those with ordinary knowledge in the field should be able to understand the meaning represented by the structure of other second chromosomes, so I will not repeat them here.
如上所述,各第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1係基於交叉機制或變異機制而產生,以下針對交叉機制及變異機制作進一步說明。 As mentioned above, each second chromosome x 1 " ( t )| t =1 ~ x 50 " ( t )| t =1 is generated based on the crossover mechanism or mutation mechanism. The following is a further description of the crossover mechanism and mutation mechanism.
在交叉機制中,資料控制中心210可隨機取得第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1的其中之二作為第一成員染色體及第二成員染色體。之後,資料控制中心210可基於第一成員染色體及第二成員染色體產生第三成員染色體,並將第三成員作為第二染色體之一而新增至第二集合Γ"(t,k)| t=1。
In the crossover mechanism, the
在一實施例中,上述第三成員染色體中對應於傳輸鏈結L (i,j)的第二能源可表徵為:
另外,上述第三成員染色體中對應於傳輸鏈結L (i,j)的第二頻段表徵為c 3,(i,j) "(t,k),且其係依據一隨機二元值而定。在一實施例中,若此隨機二元值為第一邏輯值(例如0),則c 3,(i,j)" (t,k)經定義為c 2,(i,j) '(t,k),而若隨機二元值為第二邏輯值(例如1),則c 3,(i,j) "(t,k)經定義為c 1,(i,j) '(t,k),其中c 1,(i,j) '(t,k)為第一成員染色體中對應於傳輸鏈結L (i,j)的第一頻段,c 2,(i,j) '(t,k)為第二成員染色體中對應於傳輸鏈結L (i,j)的第一頻段。簡言之,若隨機二元值為第一邏輯值,則c 3,(i,j) "(t,k)被設定為c 1,(i,j) '(t,k),而若隨機二元值為第二邏輯值,則c 3,(i,j) "(t,k)被設定為c 2,(i,j) '(t,k)。 In addition, the second frequency band corresponding to the transmission link L ( i,j ) in the third member chromosome is characterized as c 3,( i,j ) " ( t,k ), and it is based on a random binary value In one embodiment, if the random binary value is the first logical value (for example, 0), then c 3,( i,j ) " ( t,k ) is defined as c 2 , ( i,j ) ' ( t,k ), and if the random binary value is the second logical value (for example, 1), then c 3 , ( i,j ) " ( t,k ) is defined as c 1 , ( i,j ) ' ( t,k ), where c 1 , ( i,j ) ' ( t,k ) is the first frequency band in the first member chromosome corresponding to the transmission link L ( i,j ) , c 2,( i,j ) ' ( t,k ) is the first frequency band corresponding to the transmission link L ( i,j ) in the second member chromosome. In short, if the random binary value is the first logical value, then c 3,( i ,j ) " ( t,k ) is set to c 1,( i,j ) ' ( t,k ), and if the random binary value is the second logical value, then c 3,( i,j ) " ( t,k ) is set to c 2, ( i,j ) ' ( t,k ).
此外,在變異機制中,資料控制中心210可隨機取得第
一染色體x 1 '(t)| t=1~x 50 '(t)| t=1中的第四成員染色體,並隨機產生第五成員染色體。之後,資料控制中心210可基於第四成員染色體及第五成員染色體產生第六成員染色體,並將第六成員染色體作為第二染色體之一而新增至第二集合Γ"(t,k)| t=1。
In addition, in the mutation mechanism, the
在一實施例中,第六成員染色體中對應於傳輸鏈結L (i,j)的第二能源可表徵為:
另外,第六成員染色體中對應於多個傳輸鏈結L (i,j)個別的第二頻段為第四成員染色體中對應於前述傳輸鏈結L (i,j)個別的第一頻段的混洗(shuffle)版本。 In addition, the second frequency band corresponding to the multiple transmission links L ( i, j ) in the sixth member chromosome is the mixed first frequency band corresponding to the aforementioned transmission link L ( i, j ) in the fourth member chromosome. The shuffle version.
請參照圖5,其是依據本發明第一實施例繪示的產生混洗版本的示意圖。在本實施例中,假設第四成員染色體510具有對應於9個傳輸鏈結L (i,j)的第一頻段c (1,2) '(t)~c (9,10) '(t),且其個別對應的通道如圖5所示。在此情況下,資料控制中心210可從第一頻段c (1,2) '(t)~c (9,10) '(t)中隨機取出個進行混洗,其中|L|為傳輸鏈結L (i,j)的總數(即,9),.為地板函數。因此,資料控制中心210可從第一頻段c (1,2) '(t)~c (9,10) '(t)取出4個進行混洗,藉以產生第六成員染色體520中對應於各傳輸鏈結L (i,j)的第二頻段。
Please refer to FIG. 5, which is a schematic diagram of generating a shuffled version according to the first embodiment of the present invention. In the present embodiment, it is assumed
在一實施例中,資料控制中心210可重複對第一集合Γ'(t,k)| t=1執行上述交叉機制或變異機制,直至第二集合Γ"(t,k)| t=1中的第二染色體的總數達到預設數量(例如50),但本發明可不限於此。
In an embodiment, the
在取得第二集合Γ"(t,k)| t=1(其包括第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1)之後,資料控制中心210可依以下教示調整其中的一部分。
After obtaining the second set Γ" ( t,k )| t =1 (which includes the second chromosome x 1 " ( t )| t =1 ~ x 50 " ( t )| t =1 ), the
在一實施例中,資料控制中心210可基於傳輸品質限制(以γ表示)適應性地調整第二集合Γ"(t,k)| t=1中的各第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1。具體而言,資料控制中心210可取得第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1的其中之一作為第一參考染色體,其中第一參考染色體包括對應於各傳輸鏈結L (i,j)的多個第一參考能源及多個第一參考頻段。之後,資料控制中心210可計算第一參考染色體中的各第一參考頻段的傳輸品質(例如訊號雜訊干擾比(signal-to-interference-plus-noise ratio,SINR)。
In an embodiment, the
在一實施例中,對應於傳輸鏈結L (i,j)的第一參考頻段的傳輸品質可表徵如下:
之後,資料控制中心210可判斷第一參考染色體中的各
第一參考頻段是否皆滿足傳輸品質限制。亦即,資料控制中心210可判斷第一參考染色體中的各第一參考頻段的SINR是否高於γ。若是,則資料控制中心210可不調整第一參考染色體。
After that, the
另一方面,若第一參考染色體中第一參考頻段的任一者未滿足傳輸品質限制,則資料控制中心210可將第一參考頻段中之該者替換為另一頻段,以調整第一參考染色體,其中另一頻段相較於第一參考頻段之該者具有較低的負載量。
On the other hand, if any one of the first reference frequency bands in the first reference chromosome does not meet the transmission quality restriction, the
請參照圖6,其是依據本發明第一實施例繪示的調整第一參考染色體的示意圖。在本實施例中,假設第一參考染色體610包括所示的9個頻段c (1,2)(t)~c (9,10)(t),而其中共有4個頻段對應於通道1。在圖6中,假設頻段c (1,2)(t)未滿足傳輸品質限制,則資料控制中心210可相應地以另一頻段取代頻段c (1,2)(t)。在本實施例中,由於通道3僅對應於2個頻段,負載量較小,故資料控制中心210可將頻段c (1,2)(t)由對應於通道1改為對應於通道3。藉此,可產生調整後的第一參考染色體610a。
Please refer to FIG. 6, which is a schematic diagram of adjusting the first reference chromosome according to the first embodiment of the present invention. In this embodiment, it is assumed that the
接著,資料控制中心210可判斷調整後的第一參考染色體610a中的各第一參考頻段是否皆滿足傳輸品質限制。若是,則資料控制中心210可將調整後的第一參考染色體610a作為調整後的第二染色體之一而保留於第二集合中Γ"(t,k)| t=1。
Then, the
在另一實施例中,若調整的第一參考染色體610a中第一參考頻段的任一者仍未滿足傳輸品質限制,則資料控制中心210可以第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1的其中之另一取代第一參考
染色體610a。
In another embodiment, if any one of the first reference frequency bands in the adjusted
在本實施例中,假設第一參考染色體610a被取代為第一參考染色體620。之後,資料控制中心210可將第一參考染色體620中的第一參考頻段的第一特定部分與第一參考頻段中的第二特定部分一對一地對調,以調整第一參考染色體620,其中第一特定部分皆對應於第一通道,第二特定部分皆對應於第二通道,且該第一特定部分及第二特定部分具有相同的通道數量。
In this embodiment, it is assumed that the
在圖6中,第一參考染色體620中的第一特定部分例如皆對應於通道4(其通道數量為2),而第二特定部分例如皆對應於通道6(其通道數量亦為2)。在此情況下,資料控制中心210可將第一特定部分與第二特定部分一對一對調,以產生調整後的第一參考染色體620a。之後,資料控制中心210可將調整後的第一參考染色體620a作為調整後的第二染色體之一而保留於第二集合Γ"(t,k)| t=1中。
In FIG. 6, the first specific part of the
在一實施例中,在取得第一集合Γ'(t,k)| t=1(其包括第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1)及調整後的第二集合Γ"(t,k)| t=1(其包括第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1)之後,資料控制中心210可依以下教示淘汰其中的一部分,以產生對應於第1個(即,k為1)迭代操作的第三集合Γ'''(t,k)| t=1。第三集合Γ'''(t,k)| t=1包括多個第三染色體x 1 '''(t)| t=1~x 50 '''(t)| t=1,而各第三染色體x 1 '''(t)| t=1~x 50 '''(t)| t=1包括對應於各傳輸鏈結L (i,j)的第三能源及第三頻段,如圖4所示。
In one embodiment, after obtaining the first set Γ' ( t,k )| t =1 (which includes the first chromosome x 1 ' ( t )| t =1 ~ x 50 ' ( t )| t =1 ) And after the adjusted second set Γ" ( t,k )| t =1 (which includes the second chromosome x 1 " ( t )| t =1 ~ x 50 " ( t )| t =1 ), data control The
以第三染色體x 1 '''(t)| t=1為例,其可包括對應於傳輸鏈結L (1,2)至L (9,10)的第三頻段c 1,(1,2) '''( t )~ c 1,(9,10) '''(t)及第三能源P 1,(1,2) '''( t )至 P 1,(9,10) ' ''( t )。基此,本領域具通常知識者應可理解其他第三染色體的結構所代表的意義,於此不另贅述。 A third chromosome x 1 '''(t) | t = 1 , for example, which may include a transmission link corresponding to the L (1,2) to L (9,10) of the third frequency band c 1, (1, 2) ''' ( t )~ c 1,(9,10) ''' ( t ) and the third energy P 1,(1,2) ''' ( t ) to P 1,(9,10) ' '' ( t ). Based on this, those with ordinary knowledge in the field should be able to understand the meaning represented by the structure of other third chromosomes, which will not be repeated here.
在一實施例中,資料控制中心210所執行的淘汰機制如下所示。首先,資料控制中心210可取得第一染色體x 1 '(t)| t=1~x 50 '(t)| t=1及第二染色體x 1 "(t)| t=1~x 50 "(t)| t=1中的支配(dominated)染色體,並淘汰支配染色體。在不同的實施例中,資料控制中心210例如可基於一般取得支配解的方式來找出上述支配染色體,其細節於此不另贅述。
In one embodiment, the elimination mechanism implemented by the
接著,資料控制中心210可判斷淘汰後的第一染色體及第二染色體的一總數是否超過預設數量(例如50)。若否,則資料控制中心210可以第一染色體及第二染色體作為第三集合Γ'''(t,k)| t=1中的第三染色體。
Then, the
另一方面,若淘汰後的第一染色體及第二染色體的總數超過預設數量,則資料控制中心210可採用檔案庫更新(archive update)法淘汰第一染色體及第二染色體中具有較低擁擠距離的至少一者,直至剩餘的第一染色體及第二染色體的總數不超過預設數量。在本實施例中,計算擁擠距離的方式可參照先前實施例中的說明,於此不另贅述。之後,資料控制中心210可以剩餘的第一染色體及第二染色體作為第三集合Γ'''(t,k)| t=1中的第三染色體x 1 '''(t)| t=1~x 50 '''(t)| t=1。
On the other hand, if the total number of the eliminated first chromosome and second chromosome exceeds the preset number, the
基於以上教示,本領域具通常知識者應可理解在第1個時間點(即,t等於1)時,資料控制中心210如何基於初始染色體集合Γ(t,k)| t=1而產生對應於多目標最佳化演算法中的第1個迭代操作的第三集合Γ'''(t,k)| t=1。
Based on the above teachings, those with ordinary knowledge in the field should be able to understand how the
承先前所述,本發明的多目標最佳化演算法包括k max 個迭代操作,因此在第k個(1<k k max )迭代操作時,資料控制中心210可重複執行以上教示的內容,藉以產生對應於第k個迭代操作的第三集合Γ'''(t,k)| t=1。
Cheng earlier, the multi-objective optimization algorithm according to the present invention comprises a max iterations k, thus the k-th (1 <k k max ) During an iterative operation, the
具體而言,在第k個時間點時,資料控制中心210可取得第k-1個時間點對應的第三集合Γ'''(t,k-1)| t=1作為所考慮的特定染色體集合。之後,資料控制中心210即可基於此特定染色體集合依序執行比較選取法、交叉機制、變異機制、調整及淘汰機制等,以產生對應於第k個迭代操作的第三集合Γ'''(t,k)| t=1。基此,在到達迭代次數上限(即,k=k max )時,所對應的第三集合可表徵為Γ'''(t,k max )| t=1。
Specifically, at the kth time point, the
在一實施例中,在產生第1個時間點中的Γ'''(t,k max )| t=1之後,資料控制中心210可基於多個目標函數而以最佳選取(knee selection)法從Γ'''(t,k max )| t=1找出對應於第1個時間點的最佳初始染色體x *(t)| t=1。
In one embodiment, after generating Γ''' ( t, k max )| t =1 at the first time point, the
在一實施例中,上述最佳選取法例如可包括以下步驟。首先,資料控制中心210可基於第k max 個迭代操作對應的第三集合Γ'''(t,k max )| t=1的第三染色體估計關聯於能源使用率目標函數、
通道使用公平性目標函數及頻譜使用率目標函數的一最高能源使用率、一最佳通道使用公平性及一最高頻譜使用率。上述能源使用率目標函數、通道使用公平性目標函數及頻譜使用率目標函數可分別表徵如下:
之後,資料控制中心210可基於最高能源使用率、最佳通道使用公平性及最高頻譜使用率在第k max 個迭代操作對應的第三集合Γ'''(t,k max )| t=1的第三染色體中找出最佳初始染色體x *(t)| t=1,其中最佳初始染色體x *(t)| t=1具有一最低參考差值。所述最低參考差值表徵為一第一差值、一第二差值及一第三差值的總和。上述第一差值為最佳初始染色體x *(t)| t=1對應的一能源使用率與最高能源使用率之間的差值,第二差值為最佳初始染色體x *(t)| t=1對應的一通道使用公平性與最佳通道使用公平性之間的差值,而第三差值為最佳初始染色體x *(t)| t=1對應的一頻譜使用率與最高頻譜使用率之間的差值。
Thereafter, the
在一實施例中,最佳初始染色體x *(t)| t=1可表徵如下式:
在取得最佳初始染色體x *(t)| t=1之後,資料控制中心210即可依據最佳初始染色體x *(t)| t=1中各傳輸鏈結L (i,j)的初始能源及初始頻段設定各傳輸鏈結L (i,j)的最佳能源及最佳頻段。
After obtaining the best initial chromosome x * ( t )| t =1 , the
以圖4為例,假設最佳初始染色體x *(t)| t=1的內容如圖4所示,則對於傳輸鏈結L (1,2)而言,資料控制中心210可將傳輸鏈結L (1,2)的最佳能源及最佳頻段分別設定為最佳初始染色體x *(t)| t=1中對應於傳輸鏈結L (1,2)的初始能源c 11,(1,2)(t)及初始頻段P 11,(1,2)(t)。另外,對於傳輸鏈結L (9,10)而言,資料控制中心210可將傳輸鏈結L (9,10)的最佳能源及最佳頻段分別設定為最佳初始染色體x *(t)| t=1中對應於傳輸鏈結L (9,10)的初始能源c 11,(9,10)(t)及初始頻段P 11,(9,10)(t)。
Taking Figure 4 as an example, assuming that the content of the optimal initial chromosome x * ( t )| t =1 is shown in Figure 4, for the transmission link L (1,2) , the
藉此,可讓各次要用戶裝置(例如圖2中的節點1~10)在第1個時間點時能夠以最佳的能源使用率、通道使用公平性及頻譜使用率進行通訊。 In this way, each secondary user device (such as nodes 1-10 in FIG. 2) can communicate with the best energy usage rate, channel usage fairness, and spectrum usage rate at the first time point.
在第二實施例中,在其他的第t個時間點(第二實施例中
的t為大於1的正整數)時,由於已有先前時間點的歷史資料,且亦可判斷次要用戶裝置(例如圖2中的節點1~10)所處的通訊環境是否改變,故本發明可採用有別於第一實施例的初始化過程來產生可用於進行多目標最佳化演算法的染色體集合(下稱候選染色體集合)。此外,為與第1個時間點的操作產生區隔,在第二實施例中的x *(t)稱為最佳候選染色體。
In the second embodiment, at other t-th time points (t in the second embodiment is a positive integer greater than 1), since there is historical data at the previous time point, it is also possible to determine the secondary user device (For example,
請參照圖7,其是依據本發明第二實施例繪示的基於基因演算法的資源分配方法流程圖,其可由圖2的資料控制中心210執行。
Please refer to FIG. 7, which is a flowchart of a resource allocation method based on a genetic algorithm according to a second embodiment of the present invention, which can be executed by the
首先,在步驟S710中,資料控制中心210可在第t個時間點判斷次要用戶裝置(例如圖2中的節點1~10)所處的通訊環境是否改變。在一實施例中,資料控制中心210可取得所述第t個時間點的通訊環境所對應的第一可用通道數量,並取得第t-1個時間點的通訊環境所對應的第二可用通道數量。之後,資料控制中心210可判斷第一可用通道數量是否相同於第二可用通道數量。若是,則資料控制中心210可判定在第t個時間點的通訊環境未改變,反之則可判定在第t個時間點的通訊環境已改變。簡言之,只要第t個時間點的可用通道數量與第t-1個時間點的可用通道數量不同,資料控制中心210即判定次要用戶裝置所處的通訊環境已改變。
First, in step S710, the
在一實施例中,若第t個時間點的通訊環境未改變,則資料控制中心210可取得先前在第t-1個時間點對各傳輸鏈結L (i,j)分
配的歷史最佳能源及歷史最佳頻段,並據以設定各傳輸鏈結L (i,j)的最佳能源及最佳頻段。簡言之,資料控制中心210可直接沿用第t-1個時間點的資源分配結果。亦即,資料控制中心210可將對應於第t個時間點的最佳候選染色體x *(t)設定為對應於第t-1個時間點的最佳候選染色體x *(t-1),並將對應於第t個時間點的候選染色體集合Γ(t,k)設定為Γ'''(t-1,k max )(即,將Γ'''(t-1,k max )的內容全數複製至候選染色體集合Γ(t,k)中)。
In one embodiment, if the communication environment at the t-th time point has not changed, the
另一方面,在步驟S720中,反應於判定所述第t個時間點的通訊環境已改變,資料控制中心210可依據歷史候選染色體集合產生對應於第t個時間點的候選染色體集合Γ(t,k),其中候選染色體集合Γ(t,k)可包括多個候選染色體x 1(t)~x 50(t)。
On the other hand, in step S720, in response to determining that the communication environment at the t-th time point has changed, the
在一實施例中,上述歷史候選染色體集合包括對應於第t-1個時間點的第一歷史候選染色體集合Γ'''(t-1,k max )。在此情況下,資料控制中心210可將對應於第t個時間點的候選染色體集合Γ(t,k)初始化為空集合。
In an embodiment, the aforementioned set of historical candidate chromosomes includes the first set of historical candidate chromosomes Γ''' ( t -1 , kmax ) corresponding to the t-1th time point. In this case, the
之後,資料控制中心210可判斷t是否大於T,其中T為一預設時間點數量(例如10),其概念可理解為判斷目前是否具有足夠的歷史資料。若否,即代表歷史資料尚不足夠,故可採用半隨機方式建構候選染色體集合Γ(t,k),反之則可理解為歷史資料已足夠,故可採用預測方式建構候選染色體集合Γ(t,k)。
After that, the
在一實施例中,若t不大於T,則資料控制中心210可執行上述半隨機方式建構候選染色體集合Γ(t,k)。具體而言,資料控
制中心210可取得對應於所述第t-1個時間點的第一歷史候選染色體集合Γ'''(t-1,k max ),其中第一歷史候選染色體集合Γ'''(t-1,k max )包括多個第一歷史候選染色體。之後,資料控制中心210可將第一歷史候選染色體中的一部分(例如一半)複製至候選染色體集合Γ(t,k)中,以作為候選染色體的第一部分。另外,資料控制中心210還可隨機產生候選染色體的第二部分,以與候選染色體的第一部分協同建構對應於第t個時間點的候選染色體集合Γ(t,k)。
In one embodiment, if t is not greater than T, the
在另一實施例中,若t大於T,則資料控制中心210可執行上述預測方式建構候選染色體集合Γ(t,k)。在本實施例中,上述歷史候選染色體集合可更包括對應於第t-2個時間點的第二歷史候選染色體集合Γ'''(t-2,k max )。
In another embodiment, if t is greater than T, the
之後,資料控制中心210可適應性地依據第一歷史候選染色體集合Γ'''(t-1,k max )及第二歷史候選染色體集合Γ'''(t-2,k max )產生候選染色體的一第三部分。在一實施例中,候選染色體的第三部分可包括多個預測染色體,各預測染色體包括對應於傳輸鏈結L (i,j)的c (i,j)(t)及P (i,j)(t),其中c (i,j)(t)為對應於傳輸鏈結L (i,j)的預測頻段,而P (i,j)(t)為對應於L (i,j)的預測能源。
After that, the
在一實施例中,資料控制中心210產生上述第三部分的方式可如下所示。首先,資料控制中心210可建構關聯於c (i,j)(t)的一第一自迴歸模型,並據以估計關聯於c (i,j)(t)的一第一質心,其中第一質心表徵為。舉例而言,資料控制中心210可基於前幾
個時間點的c (i,j)(t)來建構上述第一自迴歸模型,而其相關細節可參照習知與自迴歸模型相關的文獻,於此不另贅述。
In an embodiment, the method for the
之後,資料控制中心210可取得c (i,j)(t-1),並依據第t-1個時間點的通訊環境將c (i,j)(t-1)映射為介於頻譜參考範圍(例如介於0及1之間)內的連續型變數。舉例而言,假設第t-1個時間點的通訊環境可表徵為相應的可用通道數量(可表示為|M(t-1)|),而c (i,j)(t-1)可屬於M(t-1)。在此情況下,映射後的c (i,j)(t-1)可表徵為,其為介於0及1之間的一連續型變數。
Afterwards, the
接著,資料控制中心210可基於第一偏移值將上述連續型變數(即,)修正為一第一副本,其中第一副本表徵為,且第一偏移值為及之間的第一歐氏距離。
Then, the
之後,資料控制中心210可基於第一質心及第一副本計算一第一頻段參考值,並判斷第一頻段參考值是否落於頻譜參考範圍(例如介於0及1之間)內。在本實施例中,第一頻段參考值可表徵為第一質心及第一副本的總和。
After that, the
在一實施例中,若第一頻段參考值落於頻譜參考範圍(例如介於0及1之間)內,則資料控制中心210可將c (i,j)(t)定義為第一頻段參考值,反之則可將c (i,j)(t)定義為一第二頻段參考值。在一實施例中,第二頻段參考值可表徵為:c (i,j)(t-1)+r 1(t)(1-c (i,j)(t-1)),其中r 1(t)為落於頻譜參考範圍內的第一隨機值。
In an embodiment, if the first frequency band reference value falls within the spectrum reference range (for example, between 0 and 1), the
另外,資料控制中心210還可建構關聯於P (i,j)(t)的一第二自迴歸模型,並據以估計關聯於P (i,j)(t)的一第二質心,其中第二質心表徵為。舉例而言,資料控制中心210可基於前幾個時間點的P (i,j)(t)來建構上述第二自迴歸模型,而其相關細節可參照習知與自迴歸模型相關的文獻,於此不另贅述。
Further,
之後,資料控制中心210可取得P (i,j)(t-1),並基於一第二偏移值予以修正為一第二副本,其中第二副本表徵為,且第二偏移值為及之間的一第二歐氏距離。
After that, the
接著,資料控制中心210可基於第二質心及第二副本計算一第一能源參考值,並判斷第一能源參考值是否落於一能源參考範圍(例如介於P min 及P max 之間)內。在本實施例中,第一能源參考值可表徵為第二質心及第二副本的總和。
Then, the
若第一能源參考值落於能源參考範圍,則資料控制中心210可將P (i,j)(t)定義為第一能源參考值,反之則可將P (i,j)(t)定義為一第二能源參考值。在一實施例中,第二能源參考值可表徵為:P (i,j)(t-1)+r 2(t)(1-P (i,j)(t-1)),其中r 2(t)為落於能源參考範圍內的一第二隨機值。
If the first energy reference value falls within the energy reference range, the
在依據以上教示產生候選染色體的第三部分之後,資料控制中心210可隨機產生候選染色體的一第四部分,以與候選染色體的第三部分協同建構對應於第t個時間點的候選染色體集合Γ(t,k)。
After generating the third part of the candidate chromosome according to the above teachings, the
在依據以上教示建構對應於第t個時間點的候選染色體 集合Γ(t,k)之後,可理解為已完成在第t個時間點的初始化操作。因此,在步驟S730中,可對候選染色體集合執行多目標最佳化演算法,以對各傳輸鏈結L (i,j)分配最佳能源及最佳頻段。 After constructing the candidate chromosome set Γ ( t, k ) corresponding to the t-th time point according to the above teachings, it can be understood that the initialization operation at the t-th time point has been completed. Therefore, in step S730, a multi-objective optimization algorithm can be performed on the set of candidate chromosomes to allocate the best energy and the best frequency band to each transmission link L ( i, j ).
在本實施例中,多目標最佳化演算法包括k max 個迭代操作,k max 為正整數,且對候選染色體集合執行多目標最佳化演算法的步驟包括:在第k個迭代操作中,對特定染色體集合採用比較選取法,以產生第一集合(1k k max );對第一集合重複執行交叉機制或變異機制,以產生第二集合;基於傳輸品質限制適應性地調整第二集合中的各第二染色體;淘汰第一集合及第二集合中的第一染色體及第二染色體的一部分,以產生對應於第k個迭代操作的第三集合,其中若k為1,則特定染色體集合為候選染色體集合,若k不為1,則特定染色體集合為第k-1個迭代操作對應的第三集合;在產生第k max 個迭代操作對應的第三集合之後,基於多個目標函數而以最佳選取法從第k max 個迭代操作對應的第三集合找出最佳候選染色體x *(t);依據最佳候選染色體x *(t)中各傳輸鏈結L (i,j)的候選能源及候選頻段設定各傳輸鏈結L (i,j)的最佳能源及最佳頻段。以上各步驟的細節可參照先前實施例中的說明,於此不另贅述。 In the present embodiment, comprises a multi-objective optimization algorithm iterations k max operation, k max is a positive integer, and the set of candidate chromosomes step multiobjective optimization algorithm comprises: k-th iteration operation , The comparative selection method is adopted for the specific chromosome set to generate the first set (1 k k max ); repeat the crossover mechanism or mutation mechanism on the first set to generate the second set; adaptively adjust each second chromosome in the second set based on the transmission quality restriction; eliminate the first set and the second set A part of the first chromosome and the second chromosome to generate the third set corresponding to the k-th iterative operation. If k is 1, the specific chromosome set is the candidate chromosome set, and if k is not 1, the specific chromosome set is a third set of the k-1 iterations corresponding to an operation; after generating a third set of k max iterations corresponding to a plurality of objective function based method to select the best k max from the operation corresponding to the third iteration Collect and find the best candidate chromosome x * ( t ); set each transmission link L ( i, j ) according to the candidate energy and candidate frequency band of each transmission link L ( i, j ) in the best candidate chromosome x * ( t) ) The best energy and best frequency band. For the details of the above steps, please refer to the description in the previous embodiment, which will not be repeated here.
此外,在一實施例中,由於最佳候選染色體x *(t)中各傳輸鏈結L (i,j)的候選頻段可能為落於上述頻譜參考範圍內的連續型變數(例如先前揭示的),故本發明的資料控制中心210可依據所述第t個時間點的通訊環境將各傳輸鏈結L (i,j)的候選頻段轉
換為離散數值。
In addition, in an embodiment, since the candidate frequency band of each transmission link L ( i,j ) in the best candidate chromosome x * ( t ) may be a continuous variable falling within the aforementioned spectrum reference range (for example, the previously disclosed ), therefore, the
舉例而言,假設第t個時間點的通訊環境(對應於M(t))對應於第一可用通道數量(可表示為|M(t)|),則資料控制中心210可將各傳輸鏈結L (i,j)的候選頻段乘以第一可用通道數量,以產生對應於各傳輸鏈結L (i,j)的參考頻段值。之後,資料控制中心210可將各傳輸鏈結L (i,j)的參考頻段值無條件進位,以產生各傳輸鏈結L (i,j)的離散數值。之後,資料控制中心210可將各傳輸鏈結L (i,j)的最佳頻段設定為對應的離散數值。另外,資料控制中心210可將各傳輸鏈結L (i,j)的最佳能源設定為對應的候選能源。
For example, assuming that the communication environment at the t-th time point (corresponding to M( t )) corresponds to the number of first available channels (which can be expressed as |M( t )|), the
簡言之,在第t個時間點所執行的多目標最佳化演算法的內容與在第1個時間點所執行的多目標最佳化演算法大致相同,惟第1個時間點所執行的多目標最佳化演算法係基於初始染色體集合Γ(t,k)| t=1而執行,但在第t個時間點所執行的多目標最佳化演算法係基於對應於第t個時間點的候選染色體集合Γ(t,k)而執行。 In short, the content of the multi-objective optimization algorithm executed at the t-th time point is roughly the same as the multi-objective optimization algorithm executed at the first time point, except that the content of the multi-objective optimization algorithm executed at the first time point is roughly the same. The multi-objective optimization algorithm of is executed based on the initial chromosome set Γ ( t,k )| t =1 , but the multi-objective optimization algorithm executed at the t-th time point is based on corresponding to the t-th The candidate chromosome set Γ ( t,k ) at the time point is executed.
之後,在步驟S740中,資料控制中心210可控制所述多個次要用戶裝置(例如圖2中的節點1~10)依據各傳輸鏈結L (i,j)對應的最佳能源及最佳頻段進行通訊。
After that, in step S740, the
由上可知,在第1個時間點之外的其他時間點中,本發明提出的基於基因演算法的資源分配方法可因應於通訊環境的改變與否而採用不同的初始化操作,進而相應地產生可用於進行多目標最佳化演算法的染色體集合候選染色體集合Γ(t,k)。在一些實施例中,若通訊環境已改變,本發明的方法還可基於先前的歷史 候選染色體集合來產生當下時間點的候選染色體集合。相較於習知僅考慮靜態通訊環境的作法,本發明係將動態通訊環境納入考量,因而能找出更符合實際情況的最佳候選染色體。並且,由於本發明係基於多目標最佳化演算法找出最佳候選染色體,因而可讓各次要用戶裝置在第t個時間點(t大於1)時能夠以最佳的能源使用率、通道使用公平性及頻譜使用率進行通訊。 It can be seen from the above that at other time points than the first time point, the resource allocation method based on the genetic algorithm proposed by the present invention can adopt different initialization operations in response to changes in the communication environment, and then generate accordingly Can be used to perform multi-objective optimization algorithm chromosome set candidate chromosome set Γ ( t, k ). In some embodiments, if the communication environment has changed, the method of the present invention can also generate the candidate chromosome set at the current time point based on the previous historical candidate chromosome set. Compared with the conventional method that only considers the static communication environment, the present invention takes the dynamic communication environment into consideration, so that the best candidate chromosomes that are more in line with the actual situation can be found. In addition, because the present invention is based on a multi-objective optimization algorithm to find the best candidate chromosomes, each secondary user device can use the best energy usage rate at the t-th time point (t is greater than 1). Channel use fairness and spectrum utilization rate for communication.
請參照圖8,其是依據本發明之一實施例繪示的資料控制中心功能方塊圖。在本實施例中,資料控制中心210可包括收發器212及處理器214。
Please refer to FIG. 8, which is a functional block diagram of a data control center according to an embodiment of the present invention. In this embodiment, the
收發器212可藉由至少包括傳送器電路、接收器電路、類比轉數位(analog-to-digital,A/D)轉換器、數位轉類比(digital-to-analog,D/A)轉換器、低雜訊放大器(low noise amplifier,LNA)、混波器、濾波器、匹配電路、傳輸線、功率放大器(power amplifier,PA)、一或多個天線單元及本地儲存媒介的組件,但不僅限於此,來為圖8的資料控制中心210提供無線存取。
The
上述接收器電路可以包括功能單元以進行如低雜訊放大、阻抗匹配、頻率混波、下頻率轉換、濾波、放大等的操作。上述傳送器電路可以包括功能單元以進行如放大、阻抗匹配、頻率混波、上頻率轉換、濾波、功率放大等的操作。A/D轉換器或D/A轉換器被配置以在上行信號處理期間轉換類比信號格式為數位信號格式,而在下行信號處理期間轉換數位信號格式為類比信號格 式。 The above-mentioned receiver circuit may include functional units to perform operations such as low-noise amplification, impedance matching, frequency mixing, down-frequency conversion, filtering, and amplification. The above-mentioned transmitter circuit may include functional units to perform operations such as amplification, impedance matching, frequency mixing, up-frequency conversion, filtering, power amplification, and the like. The A/D converter or D/A converter is configured to convert the analog signal format to a digital signal format during the upstream signal processing, and to convert the digital signal format to the analog signal format during the downstream signal processing formula.
處理器214耦接於收發器212,並可為一般用途處理器、特殊用途處理器、傳統的處理器、數位訊號處理器、多個微處理器(microprocessor)、一個或多個結合數位訊號處理器核心的微處理器、控制器、微控制器、特殊應用積體電路(Application Specific Integrated Circuit,ASIC)、現場可程式閘陣列電路(Field Programmable Gate Array,FPGA)、任何其他種類的積體電路、狀態機、基於進階精簡指令集機器(Advanced RISC Machine,ARM)的處理器以及類似品。
The
在本發明的實施例中,處理器214可藉由存取軟體模組、程式碼來實現本發明提出的基於基因演算法的資源分配方法,而其相關細節可參照先前實施例中的說明,於此不另贅述。
In an embodiment of the present invention, the
綜上所述,本發明提出的基於基因演算法的資源分配方法及資料控制中心可在所考慮的每個時間點中,先經由一定的初始化過程產生一個染色體集合,再對此染色體集合進行本發明提出的多目標最佳化演算法,以求得各傳輸鏈結所對應的最佳資源分配解,藉以讓相應的節點(即,次要用戶裝置)可據以進行傳輸。 In summary, the resource allocation method and data control center based on genetic algorithm proposed by the present invention can first generate a chromosome set through a certain initialization process at each time point under consideration, and then perform the calculation of the chromosome set. The multi-objective optimization algorithm proposed by the invention is to obtain the optimal resource allocation solution corresponding to each transmission link, so that the corresponding node (ie, the secondary user device) can perform transmission accordingly.
在第一實施例中,在第1個時間點時,由於尚未有任何歷史資料,亦無從判斷次要用戶裝置所處的通訊環境是否改變,故本發明的資料控制中心可隨機產生初始染色體集合,並據以進行多目標最佳化演算法,以找出最佳初始染色體。 In the first embodiment, at the first time point, since there is not yet any historical data, and it is impossible to determine whether the communication environment of the secondary user device has changed, the data control center of the present invention can randomly generate an initial chromosome set , And based on the multi-objective optimization algorithm to find the best initial chromosome.
在第二實施例中,在第t個時間點(t>1)時,本發明的資料控制中心則可依據通訊環境是否改變以及是否具有足夠的歷史資料來決定產生候選染色體集合的方式(例如全數複製、半隨機方式、預測方式等)。之後,本發明的資料控制中心可再對所產生的候選染色體集合執行多目標最佳化演算法,以找出對應於第t個時間點的最佳候選染色體。 In the second embodiment, at the t-th time point (t>1), the data control center of the present invention can determine the way to generate the candidate chromosome set based on whether the communication environment has changed and whether it has sufficient historical data (for example, Full copy, semi-random method, prediction method, etc.). After that, the data control center of the present invention can perform a multi-objective optimization algorithm on the generated candidate chromosome set to find the best candidate chromosome corresponding to the t-th time point.
如此一來,在考慮動態通訊環境的情況下,本發明可讓各次要用戶裝置在各個時間點時皆能夠以最佳的能源使用率、通道使用公平性及頻譜使用率進行通訊。並且,由於本發明的多目標最佳化演算法包括交叉機制、變異機制及調整機制,故可確保找出可行解,進而提升收斂效能。 In this way, considering the dynamic communication environment, the present invention allows each secondary user device to communicate with the best energy usage rate, channel usage fairness, and spectrum usage rate at each point in time. Moreover, since the multi-objective optimization algorithm of the present invention includes a crossover mechanism, a mutation mechanism, and an adjustment mechanism, it can ensure that a feasible solution is found, thereby improving the convergence performance.
雖然本發明已以實施例揭露如上,然其並非用以限定本發明,任何所屬技術領域中具有通常知識者,在不脫離本發明的精神和範圍內,當可作些許的更動與潤飾,故本發明的保護範圍當視後附的申請專利範圍所界定者為準。 Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be subject to those defined by the attached patent application scope.
S710~S740:步驟S710~S740: Steps
Claims (23)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW108142115A TWI732350B (en) | 2019-11-20 | 2019-11-20 | Resource allocation method and data control center based on genetic algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW108142115A TWI732350B (en) | 2019-11-20 | 2019-11-20 | Resource allocation method and data control center based on genetic algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
TW202121922A TW202121922A (en) | 2021-06-01 |
TWI732350B true TWI732350B (en) | 2021-07-01 |
Family
ID=77516769
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW108142115A TWI732350B (en) | 2019-11-20 | 2019-11-20 | Resource allocation method and data control center based on genetic algorithm |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWI732350B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2004068802A1 (en) * | 2003-01-30 | 2004-08-12 | University Of Surrey | Method and system for determining optimum resource allocation in a network |
WO2013170611A1 (en) * | 2012-05-17 | 2013-11-21 | 北京邮电大学 | Genetic algorithm-based scheduling and resource allocation joint optimization method |
WO2018065051A1 (en) * | 2016-10-05 | 2018-04-12 | Telefonaktiebolaget Lm Ericsson (Publ) | Controlling resource allocation in a data center |
TWI624168B (en) * | 2016-10-21 | 2018-05-11 | 元智大學 | An intelligent deployment cascade control device based on an fdd-ofdma indoor small cell in multi-user and interference environments |
WO2018228148A1 (en) * | 2017-06-15 | 2018-12-20 | 杨学志 | Wireless communication network planning method, device and system |
WO2019033749A1 (en) * | 2017-08-18 | 2019-02-21 | 中兴通讯股份有限公司 | Method, device and system for optimizing sdon architecture model, and computer readable storage medium |
WO2019127946A1 (en) * | 2017-12-26 | 2019-07-04 | 佛山科学技术学院 | Learning genetic algorithm-based multi-task and multi-resource rolling distribution method |
WO2019158034A1 (en) * | 2018-02-14 | 2019-08-22 | 华为技术有限公司 | Resource allocation method and apparatus |
-
2019
- 2019-11-20 TW TW108142115A patent/TWI732350B/en active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2004068802A1 (en) * | 2003-01-30 | 2004-08-12 | University Of Surrey | Method and system for determining optimum resource allocation in a network |
WO2013170611A1 (en) * | 2012-05-17 | 2013-11-21 | 北京邮电大学 | Genetic algorithm-based scheduling and resource allocation joint optimization method |
US9287939B2 (en) * | 2012-05-17 | 2016-03-15 | Beijing University Of Posts And Telecommunications | Method for joint optimization of schedule and resource allocation based on the genetic algorithm |
WO2018065051A1 (en) * | 2016-10-05 | 2018-04-12 | Telefonaktiebolaget Lm Ericsson (Publ) | Controlling resource allocation in a data center |
TWI624168B (en) * | 2016-10-21 | 2018-05-11 | 元智大學 | An intelligent deployment cascade control device based on an fdd-ofdma indoor small cell in multi-user and interference environments |
WO2018228148A1 (en) * | 2017-06-15 | 2018-12-20 | 杨学志 | Wireless communication network planning method, device and system |
WO2019033749A1 (en) * | 2017-08-18 | 2019-02-21 | 中兴通讯股份有限公司 | Method, device and system for optimizing sdon architecture model, and computer readable storage medium |
WO2019127946A1 (en) * | 2017-12-26 | 2019-07-04 | 佛山科学技术学院 | Learning genetic algorithm-based multi-task and multi-resource rolling distribution method |
WO2019158034A1 (en) * | 2018-02-14 | 2019-08-22 | 华为技术有限公司 | Resource allocation method and apparatus |
Also Published As
Publication number | Publication date |
---|---|
TW202121922A (en) | 2021-06-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Akkarajitsakul et al. | Distributed resource allocation in wireless networks under uncertainty and application of Bayesian game | |
CN112105062B (en) | Mobile edge computing network energy consumption minimization strategy method under time-sensitive condition | |
CN111698789B (en) | Scheduling method, apparatus and storage medium in communication system | |
Zheng et al. | Cognitive radio network duality and algorithms for utility maximization | |
Martínez-Vargas et al. | Application of NSGA-II algorithm to the spectrum assignment problem in spectrum sharing networks | |
Kang et al. | Low-complexity learning for dynamic spectrum access in multi-user multi-channel networks | |
Suman et al. | SINR pricing in non cooperative power control game for wireless ad hoc networks | |
CN104509019A (en) | Table-based link adaptation for wireless communication network transmissions | |
CN113316154A (en) | Authorized and unauthorized D2D communication resource joint intelligent distribution method | |
Chuang et al. | Dynamic multiobjective approach for power and spectrum allocation in cognitive radio networks | |
Lu et al. | Dynamic channel access and power control via deep reinforcement learning | |
He et al. | Joint optimization of channel allocation and power control for cognitive radio networks with multiple constraints | |
US20240155356A1 (en) | Resolution method for intent-based wireless network resource conflicts and apparatus thereof | |
TWI732350B (en) | Resource allocation method and data control center based on genetic algorithm | |
CN114126021A (en) | Green cognitive radio power distribution method based on deep reinforcement learning | |
TW201918109A (en) | Electronic apparatus and method for wireless communications | |
CN107613500A (en) | A kind of wireless frequency spectrum sharing method under uncertain environment | |
CN105451293B (en) | Retransmission method in wireless network, the method and apparatus for determining forwarding strategy | |
Zheng et al. | Optimal algorithms in wireless utility maximization: Proportional fairness decomposition and nonlinear Perron-Frobenius theory framework | |
Shokrnezhad et al. | Joint power control and channel assignment in uplink IoT Networks: A non-cooperative game and auction based approach | |
CN116089091A (en) | Resource allocation and task unloading method based on edge calculation of Internet of things | |
Bhattarai et al. | Improved bandwidth allocation in Cognitive Radio Networks based on game theory | |
Singh et al. | Spectrum management for cognitive radio based on genetics algorithm | |
Wang et al. | A cross-layer adaptation scheme for improving IEEE 802.11 e QoS by learning | |
Guo et al. | Deep reinforcement learning empowered joint mode selection and resource allocation for RIS-aided D2D communications |