• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    A Novel Insertion Solution for the Travelling Salesman Problem

    2024-05-25 14:43:10EmmanuelOluwatobiAsaniAderemiElishaOkeyinkaSundayAdeolaAjagbeAyodeleAriyoAdebiyiRoselineOluwaseunOgundokunTemitopeSamsonAdekunlePragasenMudaliandMatthewOlusegunAdigun
    Computers Materials&Continua 2024年4期

    Emmanuel Oluwatobi Asani ,Aderemi Elisha Okeyinka ,Sunday Adeola Ajagbe ,Ayodele Ariyo Adebiyi ,Roseline Oluwaseun Ogundokun,2,7,? ,Temitope Samson Adekunle ,Pragasen Mudali and Matthew Olusegun Adigun

    1Department of Computer Science,Landmark University,Omu Aran,251103,Nigeria

    2SDG 11 Group,Landmark University,Omu Aran,251103,Nigeria

    3Department of Computing,MiVA University,Abuja,900211,Nigeria

    4Department of Computer Science,Ibrahim Badamasi Babangida University,Lapai,911101,Nigeria

    5Department of Computer Science,University of Zululand,Kwadlangezwa,3886,South Africa

    6Department of Computer&Industrial Production Engineering,First Technical University,Ibadan,200243,Nigeria

    7Department of Multimedia Engineering,Kaunas University of Technology,Kaunas,LT-44249,Lithuania

    8Department of Computer Science,Colorado State University,Fort Collins,80523,USA

    ABSTRACT The study presents the Half Max Insertion Heuristic(HMIH)as a novel approach to solving the Travelling Salesman Problem(TSP).The goal is to outperform existing techniques such as the Farthest Insertion Heuristic(FIH)and Nearest Neighbour Heuristic(NNH).The paper discusses the limitations of current construction tour heuristics,focusing particularly on the significant margin of error in FIH.It then proposes HMIH as an alternative that minimizes the increase in tour distance and includes more nodes.HMIH improves tour quality by starting with an initial tour consisting of a‘minimum’polygon and iteratively adding nodes using our novel Half Max routine.The paper thoroughly examines and compares HMIH with FIH and NNH via rigorous testing on standard TSP benchmarks.The results indicate that HMIH consistently delivers superior performance,particularly with respect to tour cost and computational efficiency.HMIH’s tours were sometimes 16%shorter than those generated by FIH and NNH,showcasing its potential and value as a novel benchmark for TSP solutions.The study used statistical methods,including Friedman’s Non-parametric Test,to validate the performance of HMIH over FIH and NNH.This guarantees that the identified advantages are statistically significant and consistent in various situations.This comprehensive analysis emphasizes the reliability and efficiency of the heuristic,making a compelling case for its use in solving TSP issues.The research shows that,in general,HMIH fared better than FIH in all cases studied,except for a few instances(pr439,eil51,and eil101)where FIH either performed equally or slightly better than HMIH.HMIH’s efficiency is shown by its improvements in error percentage(δ)and goodness values(g)compared to FIH and NNH.In the att48 instance,HMIH had an error rate of 6.3%,whereas FIH had 14.6%and NNH had 20.9%,indicating that HMIH was closer to the optimal solution.HMIH consistently showed superior performance across many benchmarks,with lower percentage error and higher goodness values,suggesting a closer match to the optimal tour costs.This study substantially contributes to combinatorial optimization by enhancing current insertion algorithms and presenting a more efficient solution for the Travelling Salesman Problem.It also creates new possibilities for progress in heuristic design and optimization methodologies.

    KEYWORDS Nearest neighbour heuristic;farthest insertion heuristic;half max insertion heuristic;tour construction;travelling salesman problem

    1 Introduction

    Various methods exist to solve the Travelling Salesman Problem(TSP),including exact solutions and heuristics [1].Finding the shortest tour involves determining the optimal route through a set of cities,ensuring each city is visited exactly once before returning to the starting point [1–4].This completed tour is known as the Hamiltonian Cycle.It is assumed that the cost of the distance between any pair of cities is predefined.The cost often relates to distance but may represent other notions,such as time or money.A Hamiltonian cycle,as depicted in Fig.1,refers to a graph cycle that traverses all the graph’s vertices exactly once before returning to its starting vertex.The Travelling Salesman must traverse cities 1to nin a Hamiltonian cycle,that is,start from city 1,traverse the remainingn-1 cities in a specified order and then connect back to the starting city,having touched each of the cities only once at a minimal cost.

    Figure 1: A Hamiltonian weighted graph

    The TSP is simple to define,but the complexity can easily expand exponentially as the solution space increases;thus,it is classified as an NP-Hard problem[1].

    Exact techniques involve the explicit enumeration of the solution space;they try out all possible permutations of the solution.Thus,they have a complexity ofO(n!)[2–5].Exact techniques such as Dijkstra or Bellman-Ford algorithms may be deployed to effectively solve TSPs with a small degree of search space [6];others include Branch-and-Bound,Dynamic Programming Algorithms,Cutting Plane techniques and so on.Exact algorithms guarantee optimal solutions,at least hypothetically.However,the computational complexities of exact techniques are exponential [7];thus,the time required to provide their solutions grows exponentially with its solution space[6–10].Consequently,exact solutions are often impracticable and especially unsuitable for NP-hard problems with large solution space.For instance,the solution renowned as the best-performing exact technique is based on dynamic programming with a complexity ofOthus making TSPs impracticable to solve with the exact approach as the search space expands[11].These limitations of the precise method have driven the design,development,and deployment of heuristics.

    Unlike precise methods [12–15],heuristics offer rough solutions while operating within a polynomial time limit.These methods are known for their reliance on probabilities and specific rules to solve problems[16–18].For an iterative procedure,heuristics can be used when an optimal solution is guaranteed to obtain the solution quickly or make a decision within an exact process.In other words,using heuristics to solve the TSP and problems related to the TSP provides acceptable results that are not too far from the optimal yet computationally affordable.

    There are different classifications of heuristics based on the atomicity of their solution procedures,such as Tour Construction,Improvement/Local Search Heuristics,and Compound Heuristics [19–21].The Tour Construction heuristics are techniques that independently create solutions sequentially following predefined procedures within the problem space.These procedures outline the steps in the Initialization,Selection,and insertion stages.The focus of this study is on construction heuristics.Construction techniques generate reasonable approximate solutions for TSP and are equally central to the performance of the other classes of heuristics,such as improvement techniques,compound heuristics,and metaheuristics.Construction heuristics serve as a seed for the development of some heuristics and can be used to build initial solutions for high-performing techniques[22–25].Construction heuristics generally generate better initial solutions in high-performing improvement methods/metaheuristics than random initial solutions,thereby enhancing the quality of solutions[26–29].

    2 Related Works

    Construction heuristic techniques have been widely utilized in addressing traditional combinatorial optimization challenges.Various methods,such as the Nearest Neighbour Heuristic (NNH),Nearest Insertion(NIH),Cheapest Insertion,Random Insertion,Addition heuristics,Savings Heuristics,and more are commonly used.Existing tour construction methods typically fall short by between 10–30% in terms of solution quality with a worst-case complexity ofT(n)=OThe Nearest Neighbour Heuristic,for instance,is fast,flexible,and simple to implement;it,however,solves the Travelling Salesman Problem using a greedy approach and suffers immensely from the “curse of dimensionality”phenomenon[30].The Farthest Insertion,on the other hand,renowned as the bestperforming lower-order complexity heuristic[31],suffers from a high upper bound of error with farther distance[32].According to Huang et al.[32],the probability of attaining an optimal tour is higher if the distance can be reduced.

    Some well-known constructive heuristic methods are described briefly in Table 1.

    Table 1: Description of some well-known tour construction heuristics

    The Nearest Neighbour Heuristic can efficiently solve the TSP,albeit with slightly lower solution quality.The Nearest Neighbour Heuristic is a popular choice in research due to its quick implementation and straightforward approach.Experimentally,fs/fOPT≈1.26.

    Recent research has highlighted the importance of using the Nearest Neighbor Heuristic in ways such as integrating it into methods [21–23] or utilizing it as an initial step in metaheuristics to create starting solutions [7,38].While the Nearest Neighbor Heuristic is valued for its speed and simplicity,its strategy of selecting nodes with the lowest cost can lead to what is known as the“curse of dimensionality.” This means that outliers become apparent as the search space and nodes grow.The term“curse of dimensionality”is commonly used to explain how increasing dimensions result in a search space,causing data sparsity and outlier occurrences.Fischer et al.[39]described the Quadratic Traveling Salesman Problem (QTSP) as an expansion of the Traveling Salesman Problem (TSP).To solve the QTSP,they utilized a total of nine algorithms: Three exact algorithms (a polynomial transformation-based exact approach to a TSP,branch-and-bound algorithm,and branch-and-cut algorithm),seven approximate algorithms (Cheapest-Insertion Heuristic (CI),Nearest-Neighbour Heuristic(NN),Two-Directional Nearest-Neighbor Heuristic(2NN),Assignment-Patching Heuristic(AP),Nearest-Neighbour-Patching Heuristic (NNP),Two-Directional Nearest-Neighbour-Patching Heuristic (2NNP),and Greedy Heuristic (GR),and Nearest-Neighbour Heuristic (2NN) were utilized.Within around ten minutes,the branch-and-cut approach could tackle complicated real-world problems with as many as one hundred nodes in an efficient manner,successfully reaching optimality.Heuristics could handle the most complex cases in ten seconds or less,which was far faster than the running times of exact algorithms,which were acceptable.In terms of processing speed,the various iterations of the Nearest Neighbour algorithm worked well;nevertheless,when it came to the correctness of their solutions,they were not as good as the exact applications.

    Lity et al.[22]modelled the product ordering process of the incremental Software Product Line(SPL)analysis as a Travelling Salesman Problem(TSP).The aim was to optimize product orders and improve the overall SPL analysis.Products were modelled as nodes in a graph,and the solution-space information defined edge weights between product nodes.Existing graph route-finding heuristics were used to obtain the path with minimal costs.The first heuristic deployed was the Nearest Neighbour heuristic.The nodes were analyzed according to their similarity,so the NNH path was built by adding the product (node) most similar to the last node.However,it was observed that the approximation quality was poor because it first greedily added all the similar nodes and later suffered the curse of dimensionality when not-so-similar nodes were to be added.To circumvent this,a lookup was introduced to examine the next node to be added to the computed path.Thereafter,two insertion heuristics,namely Nearest Insertion and Farthest Insertion,were deployed to insert the remaining product into the existing path created by the Nearest Neighbour Heuristic.The proposed method was simulated on a prototype and evaluated for applicability and performance;a significantly more optimized SPL process was reported.

    In implementing the Iterated Local Search technique,Bernardino et al.[40] used a modified version of the Nearest Neighbor heuristic to obtain an initial answer.The Family Travelling Salesman Problem(FTSP),a famous version of the Travelling Salesman Problem(TSP),was the focus of their research efforts.The first thing they did was model the FTSP,aiming to traverse a certain number of nodes in each cluster to the lowest possible cost.After that,the FTSP sub tour was designed in both a non-compact form.Three compact models were created:The Single Commodity Flow model(SCF),the Family Commodity Flow model (FCF),and the Node Commodity Flow model (NCF),for the compact variations options such as the Connectivity Cuts(CC)model,Rounded Visits(RV)model and Rounded Family Visits(RFV)model were considered.These models were compared using experiments carried out in the C++programming language.Implementing Iterative Local Search(ILS)in C++aimed to establish limits for situations beyond what direct methods could handle.During the first phase of constructing the ILS,an initial solution was constructed using a modified version of the Nearest Neighbor heuristic.Subsequently,a local search was run to arrive at a local optimum.As a further step,they used a perturbation to break out of the local optimum,and finally,they applied removal criteria to harvest the accumulated excess nodes.A well-known research hypothesis that construction tour heuristics create excellent first solutions was verified by the International Land Survey(ILS)performance.Experiments were carried out on benchmark instances that were accessible to the general public,and the results of the experiments were recorded.Results showed that noncompact models did better than their counterpart compact ones.

    In the study by Kitjacharoenchaia [41],the Nearest Neighbour and two other heuristics were used to build an initial solution for their proposed model.Motivated by the increasing adoption of drones to achieve fast and flexible delivery,they conducted a study to simulate a drone delivery system formulated as a multiple Travelling Salesman Problem(mTSP)to minimize time.They implemented Mixed Integer Programming (MIP) to solve the problem and proposed a new technique called the Adaptive Insertion Algorithm (ADI).The ADI was implemented in two phases.An initial solution on only truck tours was built using three heuristics(namely the Nearest Neighbour Heuristic,Genetic Algorithm,and Random Cluster/tour).ThemTSPsolution was generated from the initial tour in the second phase.The method was then experimented on a single truck,multiple trucks,and a single truck and drone system,and the solution was compared with the existing MIP solution.The system reported a promising,competitive performance.It could be deduced that solutions generated from the initial solution by heuristics,such as Nearest Neighbour,hold promising performances.

    Víctor et al.[42]solved the Euclidean TSPs of small and large data sizes with an efficient heuristic that is based on the Girding Polygon,which does not take up much computer memory space and produces approximate results that are near-optimal.The computational performance of the proposed approximate heuristic was compared to that of NNH,which is another approximate heuristic.It was noticed that the proposed heuristic outperformed NNH with an average error of 16.89% while that of NN was 26.55%;it also had standard deviations of 0.05%,and NNH had 0.04%.Even though the proposed algorithm did not produce optimal solutions for the instances used,it gave an approximate solution significantly better than NNH’s.

    Insertion heuristics starts from an arbitrary point to form a sub-tour or partial circuit.Nodes not already in the sub-tour are then inserted based on predefined criteria such that the increment to the total distance of the sub-tour is minimized [23,32].Suppose that nodexis to be added to the edge(xi,xj),and given the cost functionc(xi,xj,x),then,c(xi,xj,x)=d(x,xi)+d(x,xj)-d(xi,xj).Each insertion technique method aims to add a node to an edge(that is,between two nodes)at a minimal cost.Given the sub-tourTi,and given thatxis the next node to be inserted,the insertion technique insertsxbetweenandinTi.Accordingly:

    Insertion techniques are desirable because of their speed,ease of implementation,quality of solutions,and the fact that they can be easily modified to handle complex constraints[43].There are four generally known insertion techniques:Nearest Insertion,Cheapest Insertion,Random Insertion,and Farthest Insertion.Others include Priciest Insertion,quick insertion,and greatest angle insertion[36–44].

    Insertion techniques can be used to get a good tour construction solution[45–47];according to Rosenkrantz et al.[33],insertion techniques findO(logn)approximate solutions.Insertion techniques are also used as an initial solution for improvement heuristics and metaheuristics;insertion techniques have been proven to significantly improve the performance of 2-Opt methods when used as initial solutions[46].Other researchers have presented new insertion techniques,either as a modification of state-of-the-art methods or as novel efforts.

    Experimentally,the Farthest Insertion Heuristic has been known to outperform the Random Insertion,the Cheapest Insertion,and the Nearest Insertion in that order[33–47].

    3 The Proposed HMIH Technique

    The proposed technique is an insertion method referred to in this study as the Half Max Insertion Heuristic (HMIH).The motivation was to explore some strategies with the possibility of improved tour accuracy.The design of the HMIH was motivated by two observations in literature: One,the superior solution quality of insertion techniques based on the use of polygons as an initial tour[23–42]and secondly,the limitation of the FIH’s accuracy due to the distance between its initial circuits and the next node to be inserted.Huang et al.[32]argued that although FIH performs relatively well,the distance between its circuit and new nodes to be inserted impedes its accuracy.

    The insertion heuristics randomly pick one node fromQbyinit(Q) and create a partial circuit which is expanded with every iteration.The partial circuit is made up of a minimum polygon pointu,v,w.

    LetTirepresent the partial circuit across nodes of sizeidefined asTi=(π1,π2,...,πi,π1).During the(i+1)thiteration,the insertion aims to add one node into the current circuit while minimizing the increment in the overall circuit distance.The objective is to determine how to select a node,x,fromQTiand determine how to insertxintoTito obtainTi+1.

    Consider an insertion of a nodex(/∈Ti)betweenu,vandwinTi:

    The method first determines the longest distance.dmaxof any node from either ofuorvand computeIt then find a nodewnot in the sub tour whose distance from eitheru or v≈Determine an edge(u,v)of the sub-tour to which the insertion ofwgives the smallest increase of length,i.e.,for which Δf=cux+cxv+cwx-cuvwis smallest.Insertxbetweenu,vandw.This process is iterated until a Hamiltonian cycle is formed.

    The procedure is as follows:

    The HMIH searches requireO(n)time;therefore,the time complexity of the algorithm isO(n2).The procedure is further depicted in the following flowchart in Fig.2.

    Figure 2: Flowchart of the half max insertion heuristic

    In implementing the proposed technique,the JAVA programming language version 13.0.1.was used,while GNUplot 5.2 and patch-level eight were used to plot the path graph.The heuristic was implemented on Intel Pentium Core i7 3 GHz,Windows 10(64 bit).

    4 Experimental Results

    We experimented with the HMIH,together with two State-of-the-art heuristics(namely Nearest Neighbour Heuristic (NNH) and Farthest Insertion Heuristic (FIH)) on ten publicly available benchmark instances from TSPLIB made available by Heidelberg University on http://comopt.ifi.uniheidelberg.de/software/TSPLIB95/tsp/.There were three groups of instances tested.Group one was instances whose nodes were less than 100.Group two:Instances whose nodes are more than 100 but less than 1000.Group three:Instances whose nodes are more than or equal to 1000.The implementation module generated three outputs.The first is the computation time in milliseconds(ms).Millisecond is one millionth of a second.Evidently,the accuracy of time is improved at that level of granularity.The second output is the tour cost,the distance taken to generate the tour.This is necessary for the performance evaluation of the heuristic.The third output is the tour path,the order in which the nodes join the tour.Table 2.shows the computational speed of the NNH,FIH and the proposed HMIH on the ten benchmark instances considered.Table 3 shows the tour cost of each of the three heuristics on the TSP instances.

    Table 2: The computational speed of NNH,FIH and HMIH on ten benchmark instances

    Table 3: Tour cost of NNH,FIH and HMIH on ten benchmark instances

    It is evident from Table 3 that the proposed HMIH has a shorter tour cost and is closer to the optimal tour cost in terms of solution quality than both FIH and NNH.FIH,however,compares more favourably with HMIH than NNH.

    The FIH and HMIH tour graph for some benchmark instances is presented in Figs.3–6.Fig.3 displays the path graph of FIH and HMIH for theatt48.Fig.4 shows the path graph of FIH and HMIH for theeil51.Fig.5 displays the path graph of FIH and HMIH for theeil101,and Fig.6 displays the path graph of FIH and HMIH for thech150.

    Figure 3: Path graph of FIH and HMIH for att48

    Figure 4: Path graph of FIH and HMIH for eil51

    Figure 5: Path graph of FIH and HMIH for eil101

    Figure 6: Path graph of FIH and HMIH for ch150

    5 Non-Parametric Analysis of HMIH

    Non-parametric analysis was conducted using Friedman’s test to validate the superior performance of the HMIH solutions for both NNH and FIH,which are classic tour construction heuristics.The test was conducted for all the ten instances considered in this study.The Friedman ranking was conducted based on the tour cost of the three algorithms,as presented in Table 3.The mean ranking for HMIH,NNH and FIH across all instances shown in Table 4 revealed a significant difference between the performance of HMIH and that of NNH and FIH.

    Table 4: Freidman mean ranking for the Heuristics across all instances

    These rank scores indicate that,on average,the HMIH algorithm(with the least score)performed best,followed by FIH and NNH.

    Given the rank,as presented in Table 4,Freidman’s test was then conducted using the following equation:

    where:

    nis the number of instances=10

    kis the number of TSP solutions,which in this case is 3

    Riis the sum of ranks for thei-thalgorithm

    The test was premised on the following hypothesis

    -Null Hypothesis(H0):There is no significant difference among the algorithms.

    -Alternative Hypothesis(H1):There is a significant difference among the algorithms.

    Thus,Friedman’s test on the given data instances for the three TSP heuristics(NNH,FIH,HMIH)yielded a test statistic of approximately 15.37 and ap-value of about 0.00046.Since thep-value is less than 0.05,the null hypothesis that there is no significant difference in the performance of these algorithms is therefore rejected.This indicates that the HMIH performs significantly differently from the others.

    Friedman’s analysis revealed that the HMIH technique exhibits a statistically significant superiority to NNH and FIH in performance.

    6 Performance Evaluation and Discussion of Findings

    Table 2 shows that the Nearest Neighbour Heuristic had the fastest computational speed,followed by the Farthest Insertion Heuristic and then the proposed HMI technique in all the instances.It should be noted that the proposed HMIH compared favourably with the FIH in this regard.This is consistent with literature findings that insertion techniques require more computational time than the NNH to complete tours [48–50].Additionally,the increased computational time of the proposed HMIH can be attributed to the additional computation of thehalf-maxinsertion criteria.This is consistent with works by[47–49,51],which suggest that computational speed is affected by the insertion criteria computations.

    The quality of the heuristic’s solution was assessed using the following factors:

    Percentage Error (δ):the percentage error of the heuristics’solution quality is the percentage deviation of the solution from the optimal tour solution.This is computed as×100%,wheresolηis the solution cost obtained by each heuristic,andoptis the optimal solution cost.This is equivalent to the performance ratio for suboptimal heuristics.

    Quality impr. (Σ):this involves enhancing the solution quality of the HMI method with NNH and FIH.This is computed by ENNH/FIH-EHMIH,where ENNH/FIHis the error in percentage of the NNH or FIH and EHMIHis the error in the percentage of the HMIH.

    Goodness Value( ):this is also known as accuracy.This is the inverse of error and is computed as

    Table 5 displays thepercentageerror,qualityimprandgoodnessvaluefor all the heuristics on the ten benchmark instances.

    Table 5: Percentage error,qualityimpr and goodness value for all the heuristics on the ten benchmark instances

    From Table 5,HMIH outperformed FIH in all the instances exceptpr439,eil51andeil101.FIH outperformed HMIH for pr439,while FIH and HMIH had equal tour costs foreil51andeil101.The quality of the NNH tour was,on average,24.51% lower than the quality of the highest-quality trip.Regarding the examples that were considered,the FIH average performance was 16.24%of the Held-Karp lower limit.At its highest point,the Nearest Neighbour Heuristic attained a value of 32%,while its lowest point was 17.8%.At its highest point,the Farthest Insertion Heuristic achieved a value of 26.3%,while its lowest point was 6.7%.There is a correlation between these performances and the results that have been published in the literature about NNH and FIH[47–49].Conversely,the performance of HMIH was 12.1%lower than the ideal tour duration.This was a significant difference.The HMIH that has been presented has a quality improvement of 4.14%points on average compared to the FIH.This chart,seen in Fig.7,illustrates the percentage of departure that NNH,FIH,and HMIH have from the ideal tour length.

    Figure 7: Percentage error of NNH,FIH and HMIH on the ten benchmark instances

    The shaded area of the chart denotes the quality improvement of the HMIH over the FIH.

    The proposed Half Max Insertion Heuristic consistently outperformed the Farthest Insertion.As seen by the shaded region of quality improvement in Fig.7,the heuristic was applied throughout a broad spectrum of benchmark examples,and it had a statistical significance of up to sixteen percent at one point.A comparison was made between the proposed HMIH and the Farthest Insertion,which had an average goodness value of 81.7%,and the Nearest Neighbour Heuristic,which had an average goodness value of 74.5%.This means the proposed HMIH has a higher accuracy than FIH and NNH(see Fig.8).It is worthy of note that the Farthest Insertion is considered the best-performing Insertion technique and among other lower-order complexity heuristics[31,47–50].

    Additionally,while the Farthest Insertion is faster,the computation speed of the proposed HMIH is within the same range,and since the HMIH searches were conductedO(n) times,HMIH has the same complexity ofO(n2) as the FIH and NNH.The computational speed performance of HMIH appears to follow a trend among lower-order complexity heuristics where the performing method tends to take longer computation time,perhaps owing to a more intricate process involved in getting better performance.Except for Random Insertion,which requires no computation effort to add new nodes,the better the performance,the longer the time of computation tends to be[47–52].

    Acknowledgement:The authors gratefully acknowledge the support of the Landmark University Center for Research Innovation and Development (LUCRID) for access to research repositories,literary materials,useful insights from affiliate researchers and funding.

    Funding Statement:This research is supported by the Centre of Excellence in Mobile and e-Services,the University of Zululand,Kwadlangezwa,South Africa.

    Author Contributions:The authors confirm their contribution to the paper as follows: Study conception and design: E.O.Asani,A.E.Okeyinka;data collection: A.A.Adebiyi,E.O.Asani,and R.O.Ogundokun;methodology: A.E.Okeyinka,A.A.Adebiyi,and E.O.Asani;validation and visualization:E.O.Asani,R.O.Ogundokun,T.S.Adekunle,P.Mudali,and M.O.Adigun;analysis and interpretation of results:E.O.Asani,S.A.Ajagbe,T.S.Adekunle,P.Mudali,and M.O.Adigun;draft manuscript preparation:R.O.Ogundokun,E.O.Asani,S.A.Ajagbe.All authors reviewed the results and approved the final version of the manuscript.

    Availability of Data and Materials:The data used for the implementation of this study is publicly available benchmark instances from TSPLIB made available by Heidelberg University on http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/tsp/.

    Conflicts of Interest:The authors declare that there is no conflict of interest regarding the publication of this paper.

    免费不卡黄色视频| 午夜福利乱码中文字幕| 国产精品久久电影中文字幕 | 久久久久久免费高清国产稀缺| 久久99热这里只频精品6学生| 国产精品亚洲av一区麻豆| 亚洲黑人精品在线| 男女无遮挡免费网站观看| 777久久人妻少妇嫩草av网站| 窝窝影院91人妻| 亚洲免费av在线视频| 国产福利在线免费观看视频| 90打野战视频偷拍视频| 极品少妇高潮喷水抽搐| 五月天丁香电影| 国产日韩一区二区三区精品不卡| 国产精品亚洲av一区麻豆| 国产有黄有色有爽视频| 国产91精品成人一区二区三区 | 亚洲七黄色美女视频| 国产免费av片在线观看野外av| 777久久人妻少妇嫩草av网站| 真人做人爱边吃奶动态| 欧美国产精品一级二级三级| 中文欧美无线码| 亚洲性夜色夜夜综合| 久热这里只有精品99| 在线亚洲精品国产二区图片欧美| 精品久久久精品久久久| 成人影院久久| 亚洲情色 制服丝袜| 欧美中文综合在线视频| 国产成人免费观看mmmm| 国产成人av教育| 亚洲色图 男人天堂 中文字幕| 一级片免费观看大全| 国产精品香港三级国产av潘金莲| 免费一级毛片在线播放高清视频 | 成人黄色视频免费在线看| 精品免费久久久久久久清纯 | 他把我摸到了高潮在线观看 | 免费观看a级毛片全部| 91av网站免费观看| 91麻豆精品激情在线观看国产 | 欧美日韩亚洲国产一区二区在线观看 | 国产亚洲欧美在线一区二区| 精品视频人人做人人爽| 午夜精品国产一区二区电影| 色婷婷久久久亚洲欧美| 久久青草综合色| 大片电影免费在线观看免费| 国产欧美日韩一区二区精品| 亚洲av第一区精品v没综合| 老汉色av国产亚洲站长工具| 日本欧美视频一区| 女人被躁到高潮嗷嗷叫费观| 亚洲一区中文字幕在线| 真人做人爱边吃奶动态| 窝窝影院91人妻| 日本av手机在线免费观看| 色尼玛亚洲综合影院| 精品一区二区三区四区五区乱码| 久久精品成人免费网站| 久久人妻熟女aⅴ| 久久久水蜜桃国产精品网| avwww免费| 狠狠精品人妻久久久久久综合| 国产亚洲一区二区精品| 欧美精品啪啪一区二区三区| 色综合婷婷激情| 国产亚洲午夜精品一区二区久久| bbb黄色大片| 久久免费观看电影| 成在线人永久免费视频| 久久av网站| 国产免费av片在线观看野外av| 久久久久久久大尺度免费视频| 亚洲三区欧美一区| 我要看黄色一级片免费的| 美女福利国产在线| 久久久久视频综合| 午夜老司机福利片| 视频区图区小说| 视频区图区小说| 精品熟女少妇八av免费久了| tocl精华| av在线播放免费不卡| 亚洲精品中文字幕在线视频| 在线 av 中文字幕| 一区二区三区国产精品乱码| 精品高清国产在线一区| 黄片播放在线免费| 9色porny在线观看| 久久久久久久久免费视频了| 国产免费视频播放在线视频| 久久99热这里只频精品6学生| 亚洲欧美日韩高清在线视频 | 少妇 在线观看| 12—13女人毛片做爰片一| 欧美亚洲日本最大视频资源| 在线 av 中文字幕| 91九色精品人成在线观看| 国产99久久九九免费精品| av天堂久久9| 91麻豆av在线| 国产成人免费观看mmmm| 热99re8久久精品国产| 18禁黄网站禁片午夜丰满| 在线观看免费日韩欧美大片| 电影成人av| 最近最新中文字幕大全免费视频| 精品国产一区二区三区久久久樱花| 国产亚洲欧美在线一区二区| 91精品三级在线观看| 亚洲成人国产一区在线观看| 午夜福利乱码中文字幕| 日本欧美视频一区| 精品亚洲成国产av| 久久久久久久大尺度免费视频| 欧美国产精品va在线观看不卡| 色婷婷久久久亚洲欧美| 啦啦啦在线免费观看视频4| 人人妻人人添人人爽欧美一区卜| www.熟女人妻精品国产| 欧美成人免费av一区二区三区 | 久久久欧美国产精品| 男女无遮挡免费网站观看| 国产一区有黄有色的免费视频| 久久人人97超碰香蕉20202| 超碰97精品在线观看| 亚洲成人免费电影在线观看| 老熟妇乱子伦视频在线观看| 两性午夜刺激爽爽歪歪视频在线观看 | 啪啪无遮挡十八禁网站| av有码第一页| 十八禁网站免费在线| 午夜福利,免费看| 久久毛片免费看一区二区三区| 成年动漫av网址| 国产真人三级小视频在线观看| 一个人免费看片子| tube8黄色片| 亚洲欧美色中文字幕在线| 日韩成人在线观看一区二区三区| www.自偷自拍.com| 国产97色在线日韩免费| 国产欧美日韩综合在线一区二区| 制服诱惑二区| 波多野结衣一区麻豆| 亚洲熟女精品中文字幕| 国产黄频视频在线观看| 欧美亚洲 丝袜 人妻 在线| 欧美亚洲 丝袜 人妻 在线| 在线观看免费日韩欧美大片| 国产精品亚洲av一区麻豆| 91大片在线观看| 亚洲国产欧美网| 我要看黄色一级片免费的| 在线十欧美十亚洲十日本专区| 亚洲一区中文字幕在线| 精品少妇一区二区三区视频日本电影| 变态另类成人亚洲欧美熟女 | 日韩免费高清中文字幕av| 热99国产精品久久久久久7| 日韩免费高清中文字幕av| 高清欧美精品videossex| 色老头精品视频在线观看| 欧美精品亚洲一区二区| 一边摸一边做爽爽视频免费| 亚洲成人免费电影在线观看| av天堂久久9| 精品久久蜜臀av无| 久久ye,这里只有精品| 我的亚洲天堂| 免费一级毛片在线播放高清视频 | 一本—道久久a久久精品蜜桃钙片| 一本—道久久a久久精品蜜桃钙片| 午夜激情久久久久久久| av又黄又爽大尺度在线免费看| 美女扒开内裤让男人捅视频| 国产高清videossex| xxxhd国产人妻xxx| 亚洲午夜理论影院| 一二三四在线观看免费中文在| 亚洲精品在线美女| 老司机靠b影院| 亚洲av成人不卡在线观看播放网| 国产日韩欧美亚洲二区| 国产精品亚洲一级av第二区| 免费一级毛片在线播放高清视频 | 亚洲伊人久久精品综合| 欧美成人免费av一区二区三区 | 老司机影院毛片| 一夜夜www| 免费观看人在逋| 性色av乱码一区二区三区2| 欧美日韩福利视频一区二区| 成年女人毛片免费观看观看9 | 交换朋友夫妻互换小说| 国产成人av教育| 国产男女内射视频| 黑人猛操日本美女一级片| 亚洲精品自拍成人| 亚洲av欧美aⅴ国产| 亚洲色图 男人天堂 中文字幕| 国产成人av教育| 777米奇影视久久| 最新美女视频免费是黄的| 精品卡一卡二卡四卡免费| 免费在线观看视频国产中文字幕亚洲| 亚洲色图av天堂| 美女视频免费永久观看网站| av线在线观看网站| 国产淫语在线视频| 亚洲男人天堂网一区| 一个人免费在线观看的高清视频| 久久亚洲真实| av福利片在线| 久久99热这里只频精品6学生| 欧美日韩精品网址| 伊人久久大香线蕉亚洲五| 成年动漫av网址| 日韩欧美三级三区| 国产精品自产拍在线观看55亚洲 | 国产精品免费一区二区三区在线 | 动漫黄色视频在线观看| 999久久久国产精品视频| av天堂在线播放| 无遮挡黄片免费观看| 亚洲第一青青草原| 大陆偷拍与自拍| 久久性视频一级片| 首页视频小说图片口味搜索| 一级毛片精品| 国产精品久久久久久精品电影小说| 建设人人有责人人尽责人人享有的| 亚洲国产成人一精品久久久| 亚洲国产毛片av蜜桃av| 深夜精品福利| 久久人妻福利社区极品人妻图片| 亚洲av国产av综合av卡| 性少妇av在线| 欧美精品av麻豆av| 又黄又粗又硬又大视频| 午夜两性在线视频| 免费看a级黄色片| 国产91精品成人一区二区三区 | 老司机福利观看| 精品少妇久久久久久888优播| 99九九在线精品视频| 一级黄色大片毛片| 999久久久国产精品视频| 国产不卡一卡二| 国产在线视频一区二区| 国产精品久久久久久人妻精品电影 | 久久午夜亚洲精品久久| 人人妻人人澡人人看| av一本久久久久| 免费在线观看视频国产中文字幕亚洲| 成人特级黄色片久久久久久久 | 亚洲,欧美精品.| 狠狠狠狠99中文字幕| 69精品国产乱码久久久| 夫妻午夜视频| 激情在线观看视频在线高清 | 777久久人妻少妇嫩草av网站| 黑人巨大精品欧美一区二区蜜桃| 国产精品国产av在线观看| 欧美日韩中文字幕国产精品一区二区三区 | 日日夜夜操网爽| 国产精品免费视频内射| 亚洲成人免费av在线播放| 99re6热这里在线精品视频| 欧美精品一区二区大全| 最新在线观看一区二区三区| 美女福利国产在线| 国产精品久久久久久精品古装| 考比视频在线观看| 18禁美女被吸乳视频| 亚洲人成电影免费在线| 正在播放国产对白刺激| 欧美乱码精品一区二区三区| 亚洲精品中文字幕在线视频| 99re在线观看精品视频| 国产精品.久久久| 交换朋友夫妻互换小说| 无人区码免费观看不卡 | 亚洲少妇的诱惑av| 天堂动漫精品| 久久精品91无色码中文字幕| 成人国产av品久久久| 天堂俺去俺来也www色官网| 日韩成人在线观看一区二区三区| 国产欧美日韩精品亚洲av| 69av精品久久久久久 | 在线观看免费视频日本深夜| 国产在线观看jvid| 岛国毛片在线播放| 又大又爽又粗| 99热网站在线观看| 脱女人内裤的视频| 精品少妇一区二区三区视频日本电影| 色在线成人网| av有码第一页| 汤姆久久久久久久影院中文字幕| 高清在线国产一区| 国产人伦9x9x在线观看| 老司机深夜福利视频在线观看| 999精品在线视频| 精品国产一区二区三区四区第35| 亚洲av日韩精品久久久久久密| 最新的欧美精品一区二区| 国产成人欧美在线观看 | 久久影院123| 成年版毛片免费区| 午夜福利视频在线观看免费| 国产一区二区激情短视频| 又大又爽又粗| 精品久久蜜臀av无| netflix在线观看网站| 每晚都被弄得嗷嗷叫到高潮| 久久久久精品人妻al黑| 操出白浆在线播放| 怎么达到女性高潮| 老司机在亚洲福利影院| 免费黄频网站在线观看国产| 成人免费观看视频高清| 久久午夜亚洲精品久久| 国产高清videossex| 女性被躁到高潮视频| 国产日韩欧美亚洲二区| 国产精品.久久久| 亚洲熟女精品中文字幕| 搡老乐熟女国产| 久热爱精品视频在线9| 国产淫语在线视频| 天堂中文最新版在线下载| av片东京热男人的天堂| 久久国产亚洲av麻豆专区| 久久精品国产a三级三级三级| 丝袜美腿诱惑在线| bbb黄色大片| 国产99久久九九免费精品| 日韩一区二区三区影片| 人人妻人人爽人人添夜夜欢视频| 欧美日韩成人在线一区二区| 人人澡人人妻人| 久久久精品区二区三区| 欧美激情 高清一区二区三区| 男女边摸边吃奶| 国产精品 欧美亚洲| 国产精品免费视频内射| av欧美777| 亚洲av欧美aⅴ国产| 人妻久久中文字幕网| 免费日韩欧美在线观看| 日韩制服丝袜自拍偷拍| av有码第一页| 国产精品美女特级片免费视频播放器 | 男女之事视频高清在线观看| 后天国语完整版免费观看| 黑人猛操日本美女一级片| 9热在线视频观看99| 99re在线观看精品视频| 黄频高清免费视频| 成人影院久久| 侵犯人妻中文字幕一二三四区| 精品亚洲成国产av| 露出奶头的视频| 久久天躁狠狠躁夜夜2o2o| 成年女人毛片免费观看观看9 | 母亲3免费完整高清在线观看| 亚洲免费av在线视频| 黑人操中国人逼视频| 亚洲专区中文字幕在线| 深夜精品福利| 男人操女人黄网站| 高潮久久久久久久久久久不卡| 不卡一级毛片| 国产在线免费精品| 亚洲成a人片在线一区二区| 99九九在线精品视频| 午夜福利视频在线观看免费| 男女之事视频高清在线观看| 亚洲伊人久久精品综合| 成人精品一区二区免费| 国产无遮挡羞羞视频在线观看| 十八禁网站网址无遮挡| 亚洲色图 男人天堂 中文字幕| 亚洲精品av麻豆狂野| 日韩中文字幕欧美一区二区| 久久精品人人爽人人爽视色| 可以免费在线观看a视频的电影网站| 亚洲欧美精品综合一区二区三区| 亚洲成av片中文字幕在线观看| 69精品国产乱码久久久| 啦啦啦中文免费视频观看日本| 97在线人人人人妻| 麻豆av在线久日| 日本av手机在线免费观看| 国产成人欧美| 国产一区二区三区在线臀色熟女 | 亚洲欧美激情在线| 免费在线观看完整版高清| 亚洲天堂av无毛| 中文字幕高清在线视频| 亚洲少妇的诱惑av| 51午夜福利影视在线观看| 欧美国产精品一级二级三级| 欧美成人午夜精品| www.999成人在线观看| 免费在线观看黄色视频的| 中文字幕高清在线视频| 大型黄色视频在线免费观看| 国产一区二区三区在线臀色熟女 | 国产亚洲一区二区精品| 精品人妻熟女毛片av久久网站| 在线观看免费午夜福利视频| 国产成人免费无遮挡视频| 国产亚洲精品第一综合不卡| 国产亚洲av高清不卡| 超碰成人久久| 久久精品熟女亚洲av麻豆精品| 久久久久久久久久久久大奶| netflix在线观看网站| 欧美老熟妇乱子伦牲交| 777米奇影视久久| 97人妻天天添夜夜摸| www.999成人在线观看| 热99re8久久精品国产| 免费在线观看视频国产中文字幕亚洲| 久久国产精品男人的天堂亚洲| 精品福利永久在线观看| 丰满迷人的少妇在线观看| 91字幕亚洲| 国产福利在线免费观看视频| av天堂久久9| 日韩欧美国产一区二区入口| 国产免费福利视频在线观看| 制服人妻中文乱码| 亚洲 国产 在线| 肉色欧美久久久久久久蜜桃| 高清在线国产一区| 丝袜在线中文字幕| 欧美黄色片欧美黄色片| 91国产中文字幕| 亚洲午夜理论影院| 欧美午夜高清在线| 女警被强在线播放| 欧美精品啪啪一区二区三区| 91麻豆精品激情在线观看国产 | 女同久久另类99精品国产91| 国产欧美日韩精品亚洲av| 首页视频小说图片口味搜索| 一边摸一边做爽爽视频免费| 久久久久精品人妻al黑| 亚洲成a人片在线一区二区| 女同久久另类99精品国产91| 五月开心婷婷网| 首页视频小说图片口味搜索| 亚洲五月色婷婷综合| 成人亚洲精品一区在线观看| 国产精品 欧美亚洲| 亚洲 国产 在线| 亚洲国产欧美一区二区综合| 日本av手机在线免费观看| 极品少妇高潮喷水抽搐| 久久久久国内视频| 青青草视频在线视频观看| 极品少妇高潮喷水抽搐| 高清毛片免费观看视频网站 | 精品国产超薄肉色丝袜足j| 亚洲九九香蕉| 两个人看的免费小视频| 乱人伦中国视频| 99riav亚洲国产免费| 色综合婷婷激情| 亚洲成人国产一区在线观看| 国产国语露脸激情在线看| 免费日韩欧美在线观看| 成年人免费黄色播放视频| 婷婷丁香在线五月| 亚洲国产欧美日韩在线播放| 人成视频在线观看免费观看| 99久久99久久久精品蜜桃| 日本精品一区二区三区蜜桃| 国产成人欧美在线观看 | 变态另类成人亚洲欧美熟女 | 一二三四社区在线视频社区8| 久久久精品国产亚洲av高清涩受| av免费在线观看网站| 久久中文字幕人妻熟女| 伦理电影免费视频| 天天躁狠狠躁夜夜躁狠狠躁| 成人手机av| 欧美日韩精品网址| 国产精品一区二区在线观看99| 两个人看的免费小视频| 在线观看一区二区三区激情| 69精品国产乱码久久久| 国产精品98久久久久久宅男小说| 国产男女超爽视频在线观看| 丰满迷人的少妇在线观看| 国产日韩欧美在线精品| 国产精品美女特级片免费视频播放器 | 国产在线观看jvid| 亚洲五月婷婷丁香| 97人妻天天添夜夜摸| 日韩熟女老妇一区二区性免费视频| av一本久久久久| 色在线成人网| 欧美日韩福利视频一区二区| av免费在线观看网站| 久久人妻福利社区极品人妻图片| 久久人人爽av亚洲精品天堂| 久久久国产欧美日韩av| 久9热在线精品视频| 久久精品亚洲精品国产色婷小说| 黑人巨大精品欧美一区二区蜜桃| 90打野战视频偷拍视频| 久久人妻av系列| 欧美日韩黄片免| 99在线人妻在线中文字幕 | 亚洲成a人片在线一区二区| 999久久久精品免费观看国产| 18禁美女被吸乳视频| 国产成人免费无遮挡视频| 18禁国产床啪视频网站| 老司机午夜福利在线观看视频 | 午夜福利乱码中文字幕| av片东京热男人的天堂| 午夜精品国产一区二区电影| a在线观看视频网站| 欧美人与性动交α欧美软件| 99国产精品免费福利视频| 国产亚洲午夜精品一区二区久久| 大型黄色视频在线免费观看| 成人黄色视频免费在线看| 亚洲中文字幕日韩| 18在线观看网站| 欧美日韩精品网址| 亚洲av第一区精品v没综合| 日韩人妻精品一区2区三区| 国产成+人综合+亚洲专区| 19禁男女啪啪无遮挡网站| 看免费av毛片| 老汉色av国产亚洲站长工具| 法律面前人人平等表现在哪些方面| 国产精品麻豆人妻色哟哟久久| 精品一区二区三区四区五区乱码| 在线观看www视频免费| 老鸭窝网址在线观看| 色老头精品视频在线观看| 国产伦人伦偷精品视频| 久久精品国产综合久久久| 女人久久www免费人成看片| 丰满迷人的少妇在线观看| 久久精品熟女亚洲av麻豆精品| 777米奇影视久久| 在线观看人妻少妇| 两人在一起打扑克的视频| 亚洲国产成人一精品久久久| 又大又爽又粗| 女同久久另类99精品国产91| 色精品久久人妻99蜜桃| 成年版毛片免费区| 久久亚洲真实| 亚洲欧美精品综合一区二区三区| 亚洲一区二区三区欧美精品| 变态另类成人亚洲欧美熟女 | 亚洲国产精品一区二区三区在线| av电影中文网址| 亚洲av第一区精品v没综合| 一本久久精品| 国产伦人伦偷精品视频| 午夜福利影视在线免费观看| 日韩视频一区二区在线观看| 亚洲国产欧美网| 国产一区二区三区在线臀色熟女 | 成人亚洲精品一区在线观看| 少妇猛男粗大的猛烈进出视频| 啦啦啦 在线观看视频| 50天的宝宝边吃奶边哭怎么回事| 日韩欧美三级三区| 国产一区二区 视频在线| 久久天堂一区二区三区四区| 国产人伦9x9x在线观看| 国产不卡av网站在线观看| 国产片内射在线| 1024香蕉在线观看| 久久热在线av| 中文字幕人妻丝袜一区二区| xxxhd国产人妻xxx| 啦啦啦 在线观看视频| 日日夜夜操网爽| 女人久久www免费人成看片| 啦啦啦视频在线资源免费观看| 黄色视频不卡| 99国产精品99久久久久| 国产精品一区二区在线观看99| 99久久99久久久精品蜜桃| 亚洲性夜色夜夜综合| 一本久久精品| 亚洲欧美一区二区三区黑人| 99国产精品一区二区三区| 亚洲av日韩在线播放| 下体分泌物呈黄色| 久久精品91无色码中文字幕| 午夜两性在线视频| 免费人妻精品一区二区三区视频| 日韩中文字幕欧美一区二区| 两性夫妻黄色片| 午夜视频精品福利| 热99re8久久精品国产| 精品人妻1区二区|