Congestion Control and Traffic Scheduling for ...downloads.hindawi.com/journals/wcmc/2018/9821946.pdf · Congestion Control and Traffic Scheduling for Collaborative Crowdsourcing

Research ArticleCongestion Control and Traffic Scheduling for CollaborativeCrowdsourcing in SDN Enabled Mobile Wireless Networks

Dawei Shen1 Wei Yan1 Yuhuai Peng 12 Yanhua Fu3 and Qingxu Deng 124

1College of Computer Science and Engineering Northeastern University Shenyang 110819 China2Key Laboratory of Vibration and Control of Aero-Propulsion System of Ministry of EducationNortheastern University Shenyang 110819 China3College of Jangho Architecture Northeastern University Shenyang 110819 China4State Key Laboratory of Rolling and Automation Northeastern University Shenyang 110819 China

Correspondence should be addressed to Yuhuai Peng pengyuhuaimailneueducnand Qingxu Deng dengqxmailneueducn

Received 8 September 2017 Accepted 3 January 2018 Published 21 February 2018

Academic Editor Kuan Zhang

Copyright copy 2018 Dawei Shen et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Currently a number of crowdsourcing-based mobile applications have been implemented in mobile networks and Internet ofThings (IoT) targeted at real-time services and recommendation The frequent information exchanges and data transmissions incollaborative crowdsourcing are heavily injected into the current communication networks which poses great challenges forMobileWireless Networks (MWN) This paper focuses on the traffic scheduling and load balancing problem in software-defined MWNand designs a hybrid routing forwarding scheme as well as a congestion control algorithm to achieve the feasible solutionThe trafficscheduling algorithm first sorts the tasks in an ascending order depending on the amount of tasks and then solves it using a greedyscheme In the proposed congestion control scheme the traffic assignment is first transformed into a multiknapsack problem andthen the Artificial Fish SwarmAlgorithm (AFSA) is utilized to solve this problem Numerical results on practical network topologyreveal that compared with the traditional schemes the proposed congestion control and traffic scheduling schemes can achieveload balancing reduce the probability of network congestion and improve the network throughput

1 Introduction

In recent years crowdsourcing have received extensive atten-tion from industry and academia which was originally pro-posed by American journalist Jeff Howe in 2006 Crowd-sourcing means that tasks performed by employees in a com-pany or institution before will be outsourced to the unspecificpublic networks in a free and voluntary form Tasks of crowd-sourcing are usually undertaken by the individual But if thepeople involved need to collaborate to complete the tasktheremay appear in formof individual production dependenton open source Many tasks cannot be achieved through asimple algorithm such as image labeling commodity eval-uation and entity recognition These kinds of problems aredifficult for machines to handle but can be done with crowd-sourcing In crowdsourcing it publishes tasks directly to theInternet and gathers unknown people on the Internet to solve

problems that are difficult to deal with by traditional comput-ers alone such asWikipedia reCAPTCHA[1] tagged imagesand language translations According to the different forms ofpublic participation in crowdsourcing it can be divided intocollaborative crowdsourcing and crowdsourcing contest Incollaborative crowdsourcing tasks require collaboration between the masses but people who perform tasks usually donot have rewards Crowdsourcing can effectively solve ma-chine-hard tasks by leveraging machine and a large group ofpeople on the web

Software-Defined Networking (SDN) refers to a new net-work architecture developed fromOpenFlow technology [2]SDN technology can be programmed by software to con-trol the data forwarding and ultimately achieve the purposeof free data transfer control SDN technology has outstandingadvantages in flow control therefore we hope to use SDNtechnology to solve the problem of congestion control and

HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 9821946 11 pageshttpsdoiorg10115520189821946

2 Wireless Communications and Mobile Computing

traffic scheduling [3] in crowdsourcing-based Mobile Wire-less Networks (MWN)

Currently a number of crowdsourcing-based mobile ap-plications have been applied in mobile networks and Internetof Things (IoT) targeted at real-time services and recom-mendation for example Uber Elance Amazon and AirbnbThese frequent information exchanges and data transmis-sions are heavily injected into the current communicationnetworks [4] which poses great challenges for congestioncontrol and traffic scheduling problem [5] inMobileWirelessNetworks To solve the emerging challenges this paper focus-es on the traffic scheduling and load balancing problem insoftware-defined Mobile Wireless Networks for collaborativecrowdsourcing This paper first presents a network modeltowards traffic engineering problem and then designs a hy-brid routing forwarding scheme as well as a congestion con-trol algorithm to achieve the feasible solution To validate theperformance of the proposals a lot of simulation experimentsare carried out

The rest of this paper is organized as follows Relatedwork in recent years is reviewed in Section 2 The networkmodel is then formulated in Section 3 In Section 4 designof congestion control and traffic scheduling scheme are pre-sented in detail Simulation results and analysis are discussedin Section 5 Finally conclusions are given in Section 6

2 Related Work

At present some researchers have summarized the researchwork of crowdsourcing from different perspectives

Yuen et al in [6] summarized the progress of crowdsourc-ing from applications algorithms performance and datasets Kittur et al in [7] explained the challenges of crowd-sourcing in 12 aspects such as synchronous collaborationreal-time response and dynamic machines Doan et al in[8] reviewed the crowdsourcing system applied on the worldwideweb and summarized the crowdsourcing system accord-ing to the problem type and the way of collaboration Zhaoand Zhu in [9] reviewed crowdsourcing research from fourperspectives information technology the public and orga-nization Kittur et al in [10] have studied how to decomposecomplex tasks and how to integrate workersrsquo answers to per-form initial tasks and proposed a MapReduce framework toachieve the decomposition of tasks However their method isonly suitable for specific types of tasks and the general effectis unsatisfying Scalability still needs to be solved References[11ndash13] focus on the technology of combining machine andhuman with the join operation in crowdsourcing environ-ment which first filters the problem through the machinealgorithm and then assigns the remaining problems to theworkers The authors of [12] used the transitive relationshipsof entities to further reduce the number of tasks therebysaving the cost Lofi et al in [14] reduced the cost of the taskby preprocessing data sets containing missing data throughthe ldquoerror modelrdquo and getting the answers from workersSakamoto et al in [15] studied the ways in which crowdsourc-ing participants often interact in different task types Heeret al in [16] studied how to carry out a survey and found

that the design interface wasmore suitable for crowdsourcingworkers throughquestionnairesThe authors in [17] proposeda method based on random map generation and messagingtask allocation The limitation of this method is that it canonly be used for a specific type of task to the difficulty of thetask However there are various types of task crowdsourcingplatform and some tasks need special professional knowl-edge such as language translation task Liu et al in [18] imple-mented a data analysis system to ensure the quality of theresults as themain goal first through forecastingmodel num-ber assigned tasks and then in the process of task execu-tion through online quality assessment results to determinewhether to terminate the task ahead of time thus saving costand timeThe authors in [19] proposed a new workersrsquo modelin crowdsourcing Through this model the workersrsquo qualitycan be computed accurately and timely For big data tasks thenumber of tasks affects the overall cost of the tasksThe num-ber of tasks can be reduced by effectively designing the taskthus saving the task cost Marcus et al in [20] proposed thestrategy to transform the problem of each task into multiplesubproblems But when a task contains a large number of sub-problems the price of task needs to improve Otherwise itwill be easy to cause only a small number of workers selectedtask That is to say even though such an approach reducesthe number of tasks the overall cost of the task is not guar-anteed to be reduced The authors of [21] presented a com-prehensive system model of Crowdlet that defines the taskworker arrival andworker abilitymodels In [22] the authorsdesigned an approximate task allocation algorithm that isnear optimal with polynomial-time complexity and used it asa building block to construct the whole randomized auc-tion mechanism Compared with deterministic auctionmecha-nisms the proposed randomized auction mechanism in-creases the diversity in contributing users for a given sens-ing job The authors of [23] presented a new participantrecruitment strategy for vehicle-based crowdsourcing Thisstrategy guarantees that the system can perform well usingthe currently recruited participants for a period of time in thefuture The authors in [24] focused on a more realisticscenario where users arrive one by one online in a randomorder The authors in [25] focused on the problem of howto efficiently distribute a crowdsourcing task and recruitparticipants based onD2D communications In [26] existingdefinitions of crowdsourcing were analyzed to extract com-mon elements and to establish the basic characteristics of anycrowdsourcing initiative Based on these existing definitionsan exhaustive and consistent definition for crowdsourcing ispresented and contrasted in eleven cases In [27] the authorsdefined traffic engineering as a large-scale network project tosolve the performance evaluation and network optimizationin the network In [28] traffic engineering has been furtherexplained and the traffic engineering is a route optimizationmethod to improve the quality of network service by avoidingthe link congestion in the network

3 Network Model

There are a number of possible next hops thatmay occur afterthe crowdsourcing task has selected the assignment object in

Wireless Communications and Mobile Computing 3

Next hop 1

Next hop 2

XY loadbandwidth

Task source nodeload 20

Task destination node

10100 40100

20100 20100

Figure 1 The illustration of the next hop selection in our network model

the mobile wireless network and different next hop optionsaffect the load balancing in the network As shown in Figure 1if the crowdsourcing task source node in the figure forwardsthe mission to the destination node through the next hop 1the maximum link utilization rate in the network is 06 If thenext hop 2 is chosen to forward the task assignment then themaximum link utilization in the network is 04Therefore theSDN controller in the network needs to calculate the next hopperiodically to achieve the load balancing in the network

Taking into account the fact that the routing networkalready exists in the current mobile wireless network itrequires a lot of manpower and resources to replace all thewireless nodes for the SDN node [29]Therefore we considerthe SDN node in themobile wireless network part of the con-figuration of the scenarioWe assume the nodes in themobilewireless network run the OSPF protocol so the SDN con-troller can collect the load information of the links in the net-work And the SDN nodes can obtain the link utilization rateof all the links in the network When the crowdsourcing taskleaves the SDN node it may pass through other SDN nodeson its forwarding path These nodes can also have multiplenext hops as shown in Figure 2 where the yellow nodes rep-resent SDN nodes while the white nodes represent non-SDNnodes In addition the solid line represents the forwardingpath and the dotted line represents the possible forwardingpaths It is assumed that the current task node 1 is an SDNnode and it selects node 2 as its next hopThere is also anoth-er SDN node 4 on the forwarding path and the next hop ofnode 4 may be node 5 or node 6 Through the coordinationof multiple SDN nodes distributed in the Mobile WirelessNetworks we can have multiple possible forwarding paths tocarry the crowdsourcing tasks to achieve load balancing forglobal networks

Therefore we first need to find out all the possible pathsthat the package task forwards We use the tree structure tobuild all possible forwarding paths [30] First we constructthe source node of the task as the root of the tree Each node inthe tree can be divided into SDN nodes and non-SDN nodesIf it is an SDN node then it can have multiple child nodesotherwise it only has a child node We assume that whenthe package task is forwarded each node in the networkwill inject an identity packet of the current node Whenpassing through the SDN node we check this identity packetand remove the branch path containing the nodes thatalready exist in the current identity packet ensuring that the

loopback is not generated when the packet task is forwardedIn Figure 2 for example the tree structure of all possibleforwarding paths can be constructed as shown in Figure 3

In what we described above there is only one crowd-sourcing task in the wireless sensor network but in realitythere can be multiple crowdsourcing task in the network [31]

Formally the wireless sensor network can be modeledas 119866(119881 119864) with the node set 119881(1 2 V) and link set119864(11989012 11989013 119890119894119895)

Assume that there is no interference between nodes andlinks Suppose 119879 is the crowdsourcing tasks matrix and thetask set is119873(112 2

12 119899

119904119889) (119904 is the crowdsourcing task source

node and 119889 is the crowdsourcing task destination node) Andthe amount of task is 119871(119899119904119889) Define 119862(119890) as the link capacityDefine link utilization as119880(119890) which can be formulated as in

119880 (119890) =sum119890 119871 (119899

119904119889)

119862 (119890) (1)

Define that when a crowdsourcing task passes througha node all possible forwarding paths are added to set 119875119899

119904

119889

V There are two scenarios in themobile wireless network whena number of crowdsourcing tasks pass through non-SDNnodes they can only be forwarded in accordance with theOSPF protocol to its next hop And when multiple crowd-sourcing tasks pass through SDN nodes we have multiplepossible forwarding paths In the mobile wireless networkwe can only control SDN nodes Therefore when the crowd-sourcing task traffic passes through the SDN node the prob-lemwe need to solve is as follows given119866119873119862(119890) and119875 howwe schedule the task119873 over the path 119875with the path capacity119862(119890) to minimize the maximum link utilization 119880 thenachieving load balancing We describe it as problem 119878(119873119871(119899119904119889) 119862(119890) 119875 119880)

Given the definitions above the problem can be formal-ized as follows

Minimize 119880 (2)

Subject to sum119890

119871 (119899119904119889) le 119880 sdot 119862 (119890) forall119890 isin 119864 (3)

119875119871(119899119904

119889) ge 0 forall119875 (4)

119871 (119899119904119889) ge 0 forall119899119904119889 isin 119873 (5)

Formula (3) indicates that the size of the task on any linkis less than or equal to the maximum link utilization in the


1

2

3

4

5

6Task source node


Figure 2 Multipath selection illustration

1

2 3

4 4

5 6 5 6

6 6

Task source node





Figure 3 All possible path tree diagram

networkmultiplied by the link capacity Formula (4) indicatesthat the amount of task on any forwarding path should benonnegative Formula (5) indicates that task should be non-negative

4 Congestion Control andTraffic Scheduling Schemes

41 Design of Hybrid Routing and Forwarding Algorithm Inour model we divided the nodes in the mobile wireless net-work into two categories SDN nodes and non-SDN nodesWhen the crowdsourcing task traffic passes through the non-SDN node we use OSPF protocol to perform the next hoprouting When the crowdsourcing task traffic passes throughthe SDN node we describe this as problem 119878(119873 119871(119899119904119889) 119862(119890)119875 119880) There is a special case of problem 119878 where 119875 = 1and 119880 = 1 The problem 119878(119873 119871(119899119904119889) 119862(119890) 1 1) is NP and wecan reduce the well-known 0-1 knapsack problem [32] to thisproblem Therefore 119878(119873 119871(119899119878119889) 119862(119890) 1 1) is NP-hard Thusthe more general problem 119878(119873 119871(119899119904119889) 119862(119890) 119875 119880) is also NP-hard This means the computation cannot be completed ina reasonable time for large networks Therefore we developa heuristic algorithm for this problem with polynomial-timecomplexity

On the algorithm we make the following assumptions(1) SDN control center can be aware of the relevant

information in the network correctly and timely(2) Network topology is stable in a short time and we do

not consider the interference of wireless networks

(3) All the nodes are running standard OSPF protocolnodes in the mobile wireless network in addition toSDN nodes

(4) Mobile wireless network has only one SDN controller

(5) In the process of routing SDN nodes select only onepath to forward when processing a crowdsourcingtask flow

(6) The task flow is forwarded hop by hop

In this case we assume that none of the links in thenetwork will be congested and there will not be a number ofcrowdsourcing task traffic on a link exceeding the capacity ofthe linkTherefore when the SDN node forwards the crowd-sourcing task we can sort the crowdsourcing tasks accordingto the task loadThen according to the greedy algorithm thecrowdsourcing task is distributed to the corresponding linkwhich makes the value of maximum link utilization in thenetwork minimum

The hybrid routing and forwarding algorithm is given inAlgorithm 1

Since we define the utilization of the link as the ratioof the link capacity of the data flow on the current link ifthe data flow is far greater than our link capacity our linkutilization will be greater than 1 So the networkrsquos maximumlink utilization is greater than 1 which is contrary to the ideaof load balancing in traffic engineeringTherefore our crowd-sourcing task trafficmatrix cannot be generated arbitrarily as


Algorithm for hybrid routing and forwarding(1) Begin(2) Input mobile wireless network topology graph 119866(119881 119864) crowdsourcing task flow matrix 119879(3) for each row in 119879 do(4) If V is non-SDN node then(5) Assign the task flow to its next hop forwarding link(6) repeat(7) forallV isin 119881 V + +(8) until all non-SDN nodes are traversed(9) If V is SDN node then(10) Sort the task in ascending order according to the load of the task flow(11) Compute all possible forwarding path 119875(12) Use the greedy algorithm to assign task to its next hop forwarding link(13) repeat(14) forallV isin 119881 V + +(15) until all SDN nodes are traversed(16) Compute link utilization on all links in the network Get the maximum link utilization 119880(17) Update the crowdsourcing task traffic matrix 119879 to 1198791015840(18) If 1198801015840 ge 119880 then(19) 119880 larr 1198801015840(20) Return to the third step(21)Output the maximum link utilization 119880(22) End

Algorithm 1 Hybrid routing and forwarding algorithm

for the task flow size according to the method described inliterature [33] we generate the formula as follows

119889119894119895 = 120590119894 sum119905|(119894119905)isin119864

119888(119894119905)sum119905|(119905119895)isin119864 119888(119905119895)

sum(119898119899)|(119898119899)isin119864 119888(119898119899) minus sum119905|(119894119905)isin119864 119888(119894119905)

119894 119895 isin 119881

(6)

In formula (6) 119889119894119895 represents the size of the traffic flowfrom the source node 119894 to the destination node 119895 120590119894 representsa random number in an interval [0 1] 119888(119894 119905) represents thelink capacity between the source node 119894 and its neighboringnode 119905 119888(119905 119895) is the link capacity between destination node 119895and its neighboring node 119905 and 119888(119898 119899) represents the capac-ity on the link (119898 119899) We generate 40 sets of crowdsourcingtask flow matrices as simulation data according to formula(6) According to the above conditions we have simulated theproposed algorithm

42 Design of Congestion Control Algorithm As mentionedabove we assume that there will be no congestion in themobile wireless network but in fact congestion is inevitablein the process of mass crowdsourcing Therefore the prob-lem 119878(119873 119871(119899119904119889) 119862(119890) 119875 119880) should be 119878(119873 119871(119899119904119889) 119862(119890) 119875119890 1)because the maximum utilization of the link is 1 and 119875119890 isthe first link of the possible path 119875 In this case when anSDN node is forwarding the crowdsourcing task it needs to

select a subset of its task set 119873 1 2 119899 first Then thesesubtasks will be assigned to the possible forwarding link 119875119890with the maximum value of assigned tasks under the limi-tation of each link It is a multiknapsack problem MultipleKnapsack Problem (MKP) refers to the selection of a subsetof items in an item collection 119873 1 2 119899 to be loadedinto 119872 1 2 119898 backpack The purpose is to maximizethe total value of selected items with the total capacity notexceeding the volume of each backpack Here we use theAFSA algorithm in [34] to solve this problem Artificial FishSwarm Algorithm (AFSA) is a new intelligent optimizationalgorithm for biomimetic group Artificial fish can makeAFSA better intelligent and suitable for solving large-scalecomplex optimization problems We assign the crowdsourc-ing tasks as many as possible to the link without exceedingthe link capacity According to this heuristic rule if we wantto assign the task 119894 to the link 119895 there are two possibilitiesOne is the link capacity 119862(119895) lt 119871(119894) and we cannot assignthe task to the link The other one is the link capacity 119862(119895) ge119871(119894) Let 119862119903(119890) represent the remaining capacity of the link 119890There are two conditions (1) 119862119903(119895) ge 119871(119894) if task 119894 is neverassigned to any link then task 119894 is assigned to the link 119895 and119862119903(119895) = 119862119903(119895) minus 119871(119894) if task 119894 was assigned to link 119896 (119896 = 119895)we firstly execute TakeOut(119894 119896) (TakeOut(119894 119896) which meanstaking the task 119894 out of link 119896 and then 119862119903(119896) = 119862119903(119896) + 119871(119894))Then we assign the task 119894 to the link 119895 and the remainingcapacity of the link 119895 decreases 119871(119894) (2) 119862119903(119895) lt 119871(119894) we


Algorithm for Congestion control(1) Begin(2) Input mobile wireless network topology graph 119866(119881 119864) crowdsourcing task flow matrix 119879(3) for each row in 119879 do(4) If V is non-SDN node then(5) Assign the task flow to its next hop forwarding link(6) repeat(7) forallV isin 119881 V + +(8) until all non-SDN nodes are traversed(9) If V is SDN node then(10) Compute all possible forwarding path 119875119890(11) Compute the link capacity 119862119903(119875119890)(12) Use the AFSA algorithm to assign task flow to its next hop forwarding link(13) repeat(14) forallV isin 119881 V + +(15) until all SDN nodes are traversed(16) Update the crowdsourcing task traffic matrix 119879 to 1198791015840 119879119898 = 119879119898 + 1 Return to the third step(17) Output Number of times 119879119898(18) End

Algorithm 2 Congestion control algorithm

executeTakeOut(119901 119895) (119901 is any task that is assigned to the link119895) until 119862119903(119895) ge 119871(119894) and then we execute (1) The artificialfish is always kept in a feasible solution and close to the boundboundary The effective optimization of artificial fish underthe guidance of behavior strategy was carried out by artificialfish feeding rear-ending and clustering

Since we can only control SDN nodes in the network wewill take the traffic of non-SDN nodes in the forwarding linkfirst The remaining capacity of the link is the backpackingcapacity of our multibackpack problem We also need toassume that the crowdsourcing task flow cannot be splitAssume that when the number of tasks on a link exceeds thelink capacity of the link it causes the task to be discarded andneeds to be reposted Define 119879119898 as the number of times thatthe crowdsourcing task has been forwarded Finally we eval-uate our congestion control algorithm by calculating the linkthroughput We use formula (7) to compute the throughputof the network

Throughput = sum119871 (119899)119879119898

(7)

The congestion control algorithm is given in Algorithm 2

5 Number Results and Analysis

We mainly use VS2010 to complete the simulation which iscoded in CC++We use the wireless network standard basedon IEEE 80211b [35] to build our mobile wireless networkwith a maximum bandwidth of 11Mbps which means themaximum link capacity can be set to 11M Here we use

the method described in [36] to set the link capacity inmobile wireless network First divide all the nodes into twocategories according to the degree of each node A class noderepresents those nodes whose degree is less than 3 and B classrepresents the set of other degrees of nodes If a link has twonodes in the B class node set then the link capacity is 11Mif there is a node in the link in the A class node set the linkcapacity of 6M

Simulation of the mobile wireless network topologies areshown in Figures 4 and 5 where yellow nodes represent SDNnodes and white nodes represent non-SDN nodes and wesimulate the experiment by increasing the number of SDNnodes gradually

For the hybrid routing and forwarding algorithm wecompare the network with no SDN nodes by increasing theamount of SDN nodes in the network which is the networkthat we assume all nodes are forwarded according to theOSPF protocol From (a) to (d) compare the maximum linkutilization between our proposed hybrid routing forwardingscheme and the OSPF protocol by increasing the number ofSDNnodesThe simulation results are shown in Figures 6 and7

Figures 6 and 7 present analyses of the maximum utiliza-tion with different SDN nodes deployment in Topology 1 andTopology 2 The simulation results are shown in Figures 6and 7 the 119910-axis represents the maximum link utilizationand the 119909-axis represents the number of crowdsourcing taskflowmatrixes We can see intuitively that with the increase inSDN nodes the overall trend of maximum link utilization isdecreasing in the mobile wireless network from the simula-tion results in Figures 6 and 7 However it can be seen in the


5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes

Figure 4 Mobile wireless network topology 1

0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes


figure that when the SDN nodes in the network are relativelyfew the maximum link utilization in the network obtainedfrom the hybrid routing and forwarding algorithm is almostthe same compared with the OSPF routing algorithmThis isbecause the SDN controller can only control the SDN nodesto manipulate the traffic in the network When the SDNnodes in the network are few the traffic in the whole networkbecomes uncontrollable Although traffic through the SDN

nodes can be controlled the maximum utilization rate of thelocal link in the network is reduced and the local networkcan achieve load balancing and it is difficult to achieve loadbalancing for the whole network In addition by comparingTopology 1 and Topology 2 the benefits of deploying SDNnodes will become more apparent as the number of nodesin the network increases and the network topology becomesmore complex


0010203040506070809

1Th

e max

imum

link

util

izat

ion

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 201Crowdsourcing task flow matrix serial number

Hybrid routing forwarding algorithmOSPF

(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


OSPFHybrid routing forwarding algorithm

(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes

Figure 6 Comparison of the maximum utilization with different SDN nodes deployment in Topology 1

For the congestion control algorithm we gradually in-crease the number of SDN nodes in the mobile wireless net-work We calculate the throughput of the network throughformula (7) and then compare it The simulation results areshown in Figure 8

Figure 8 shows the analysis of throughput for differentSDN nodes deployment in Topology 1 and Topology 2 It canbe observed that the throughput performance of Topology 1and Topology 2 are both better with SDN nodes increasingFrom the comparison results in Figure 8 it can be concludedthat our congestion control algorithm can effectively improvethe network throughput

6 Conclusion

At present massive crowdsourcing-based mobile applica-tions have been applied in mobile networks and IoT targetedat real-time services and recommendation The frequentinformation exchanges and data transmissions in collabora-tive crowdsourcing are continually injected into the currentcommunication networks which poses great challenges inMobileWirelessNetworks (MWN)This paper focuses on thetraffic scheduling and load balancing problem in software-defined MWN and designs a greedy heuristic algorithm aswell as a congestion control algorithm to achieve feasiblesolutionsThe proposed traffic scheduling algorithm sorts the

tasks in ascending order according to the amount of tasks andthen solves them using the greedy scheme The packet taskis assigned to the corresponding link for forwarding so thatthe maximum link utilization in MWN is the least In theproposed congestion control scheme the traffic assignmentis transformed into a multiknapsack problem and then theAFSA algorithm is employed to solve the problem The nodeselects a subset in its feasible task set and assigns it to thep links whichmakes themaximum amount of tasks allocatedwithout exceeding the limited capacity of each link Thesimulation results demonstrate that compared with the tra-ditional schemes the proposed congestion control and trafficscheduling methods can achieve load balancing reduce theprobability of network congestion and improve the networkthroughput

Conflicts of Interest

The authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

This work was supported in part by the National NaturalScience Foundation ofChina (61501105) theNationalKeyRe-searchandDevelopmentProgramofChina (2016YFC0801607)


0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes


1-SDN node 3-SDN node 6-SDN node ALL-SDN node

Topology 1Topology 2

0

50

100

150

200

250

300

350

Thro

ughp

ut

Figure 8 Comparison of throughput with different SDN nodes deployment

the Fundamental Research Funds for the Central Universities(N150404018 N130304001 N150401002 and N161608001)and the Open Research Fund from the State Key Laboratoryof Rolling and Automation Northeastern University Grantno 2017RALKFKT002

References

[1] L vonAhn BMaurer CMcMillen D Abraham andM BlumldquoreCAPTCHA human-based character recognition via web

security measuresrdquo American Association for the Advancementof Science Science vol 321 no 5895 pp 1465ndash1468 2008

[2] N McKeown T Anderson H Balakrishnan et al ldquoOpenFlowenabling innovation in campus networksrdquoACMSigcommCom-puter Communication Review vol 38 no 2 pp 69ndash74 2008

[3] RWang D Butnariu and J Rexford ldquoOpen Flow-based serverload balancing gone wildrdquo in Proceedings of The 11Th USENIXConference on Hot Topics in Management of Internet CloudAnd Enterprise Networks And Services pp 12-12 USENIXAssociation 2011


[4] XKongX Song F XiaHGuo JWang andA Tolba ldquoLoTADlong-term traffic anomaly detection based on crowdsourced bustrajectory datardquoWorldWideWeb Information Systems pp 1ndash232017

[5] Z Ning F Xia N Ullah et al ldquoVehicular social networksenabling smart mobilityrdquo IEEE Communications Magazine vol55 no 5 pp 16ndash55 2017

[6] M C Yuen I King and K S Leung ldquoA Survey of crowdsourc-ing systemsrdquo in Proceedings of the IEEEThird International Con-ference on Social Computing (SocialCom) pp 766ndash773 IEEEBoston MA USA October 2012

[7] A Kittur J V Nickerson M Bernstein et al ldquoThe future ofcrowd workrdquo in Proceedings of the 2013 Conference on Computersupported cooperative work vol 3-4 pp 1301ndash1318 Social Sci-ence Electronic Publishing San Antonio Texas USA Feburary2013

[8] A Doan R Ramakrishnan and A Y Halevy ldquoCrowdsourcingsystems on the world-wide webrdquo Communications of the ACMvol 54 no 4 pp 86ndash96 2011

[9] Y Zhao and Q Zhu ldquoEvaluation on crowdsourcing researchCurrent status and future directionrdquo Information Systems Fron-tiers vol 16 no 3 pp 417ndash434 2014

[10] A Kittur B Smus S Khamkar and R E Kraut ldquoCrowd-Forge Crowdsourcing complex workrdquo in Proceedings of the24th Annual ACM Symposium on User Interface Software andTechnology UISTrsquo11 pp 43ndash52 USA October 2011

[11] JWang T KraskaM J Franklin et al ldquoCrowdER crowdsourc-ing entity resolutionrdquo Proceedings of the VLD Endowment vol5 no 11 pp 1483ndash1494

[12] JWangG Li T KraskaM J Franklin and J Feng ldquoLeveragingtransitive relations for crowdsourced joinsrdquo in Proceedings ofthe 2013 ACM SIGMOD Conference on Management of DataSIGMOD 2013 pp 229ndash240 USA June 2013

[13] G Demartini D E Difallah and P Cudre-Mauroux ldquoZen-Crowd Leveraging probabilistic reasoning and crowdsourcingtechniques for large-scale entity linkingrdquo in Proceedings of the21st Annual Conference onWorldWideWebWWWrsquo12 pp 469ndash478 France April 2012

[14] C Lofi K El Maarry andW-T Balke ldquoSkyline queries over in-complete data - Error models for focused crowd-sourcingrdquo Lec-tureNotes in Computer Science (including subseries LectureNotesin Artificial Intelligence and Lecture Notes in Bioinformatics)Preface vol 8217 pp 298ndash312 2013

[15] Y Sakamoto Y Tanaka L Yu and J V Nickerson ldquoThe Crowd-sourcing design spacerdquo in Proceedings of the International Con-ference on Foundations of Augmented Cognition Directing theFuture of Adaptive Systems vol 6780 of Lecture Notes inComputer Science pp 346ndash355 Springer-Verlag 2011

[16] J Heer and M Bostock ldquoCrowdsourcing graphical perceptionUsing mechanical Turk to assess visualization designrdquo in Pro-ceedings of the 28th Annual CHI Conference on Human Factorsin Computing Systems CHI 2010 pp 203ndash212 USA April 2010

[17] D Karger R S Oh and D Shah ldquoIterative learning for reliablecrowdsourcing systemsrdquo in Proceedings of the InternationalConference on Neural Information Processing Systems pp 1953ndash1961 Curran Associates Inc 2011

[18] X Liu M Lu B C Ooi et al ldquoCDAS A crowdsourcing dataanalytics systemrdquo Proceedings of the VLDB Endowment vol 5no 10 pp 1040ndash1051 2012

[19] J Feng G Li H Wang et al ldquoIncremental quality inference incrowdsourcingrdquo in Proceedings of the International Conference

on Database Systems for Advanced Applications Lecture Notesin Computer Science pp 453ndash467 Springer International Pub-lishing 2014

[20] A Marcus E Wu D Karger S Madden and R Miller ldquoHu-man-powered sorts and joinsrdquo Proceedings of the VLDB Endow-ment vol 5 no 1 pp 13ndash24 2011

[21] L Pu X Chen J Xu and X Fu ldquoCrowdlet Optimal worker re-cruitment for self-organized mobile crowdsourcingrdquo in Pro-ceedings of the 35th Annual IEEE International Conference onComputer Communications IEEE INFOCOM 2016 USA April2016

[22] J Li Y Zhu Y Hua and J Yu ldquoCrowdsourcing sensing tosmartphones A randomized auction approachrdquo in Proceedingsof the 23rd IEEE International Symposium on Quality of ServiceIWQoS 2015 pp 219ndash224 USA June 2015

[23] Z He J Cao and X Liu ldquoHigh quality participant recruitmentin vehicle-based crowdsourcing using predictable mobilityrdquo inProceedings of the 34th IEEE Annual Conference on ComputerCommunications and Networks IEEE INFOCOM 2015 pp2542ndash2550 May 2015

[24] D Zhao X-Y Li and H Ma ldquoHow to crowdsource taskstruthfully without sacrificing utility Online incentive mecha-nisms with budget constraintrdquo in Proceedings of the 33rd IEEEConference on Computer Communications IEEE INFOCOM2014 pp 1213ndash1221 May 2014

[25] Y Han and H Wu ldquoMinimum-Cost Crowdsourcing with Cov-erageGuarantee inMobileOpportunisticD2DNetworksrdquo IEEETransactions on Mobile Computing vol 16 no 10 pp 2806ndash2818 2017

[26] E Estelles-Arolas and F Gonzalez-Ladron-De-Guevara ldquoTo-wards an integrated crowdsourcing definitionrdquo Journal of Infor-mation Science vol 38 no 2 pp 189ndash200 2012

[27] Y Lee and B Mukherjee ldquoTraffic engineering in next-gene-ration optical networksrdquo Communications Surveys amp TutorialsIEEE vol 6 no 3 pp 16ndash33

[28] DAwduche A Chiu A Elwalid IWidjaja andX Xiao ldquoOver-view and principles of internet traffic engineeringrdquo IETF RFC3272 Academy of Science Engineering and Technology 2002

[29] Z Ning Q Song Y Yu Y Lv X Wang and X Kong ldquoEner-gy-aware cooperative and distributed channel estimationschemes for wireless sensor networksrdquo International Journal ofCommunication Systems vol 30 no 5 Article ID e3074 2017

[30] C-Y Chu K Xi M Luo and H J Chao ldquoCongestion-awaresingle link failure recovery in hybrid SDN networksrdquo in Pro-ceedings of the 34th IEEE Annual Conference on Computer Com-munications and Networks IEEE INFOCOM 2015 pp 1086ndash1094 Hong Kong May 2015

[31] Z Ning X Hu Z Chen et al ldquoA cooperative quality-awareservice access system for social internet of vehiclesrdquo IEEEInternet of Things Journal pp 1-1

[32] M R Garey and D S Johnson Computers and Intractability AGuide to the Theory of NP-Completeness W H Freeman SanFrancisco Calif USA 1979

[33] K Li S Wang S Xu and X Wang ldquoERMAO An enhancedintradomain traffic engineering approach in LISP-capable net-worksrdquo in Proceedings of the 54th Annual IEEE Global Telecom-munications Conference ldquoEnergizing Global CommunicationsrdquoGLOBECOM 2011 USA December 2011

[34] X MA and Q LIU ldquoArtificial fish swarm algorithm for multipleknapsack problemrdquo Journal of Computer Applications vol 30no 2 pp 469ndash471 2010


[35] Z Ning F Xia N Ullah X Kong and X Hu ldquoVehicular SocialNetworks Enabling Smart Mobilityrdquo IEEE CommunicationsMagazine vol 55 no 5 pp 16ndash55 2017

[36] H Yang IP routing based on topology of network traffic engineer-ing research University of electronic science and technology2013

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


traffic scheduling [3] in crowdsourcing-based Mobile Wire-less Networks (MWN)

Currently a number of crowdsourcing-based mobile ap-plications have been applied in mobile networks and Internetof Things (IoT) targeted at real-time services and recom-mendation for example Uber Elance Amazon and AirbnbThese frequent information exchanges and data transmis-sions are heavily injected into the current communicationnetworks [4] which poses great challenges for congestioncontrol and traffic scheduling problem [5] inMobileWirelessNetworks To solve the emerging challenges this paper focus-es on the traffic scheduling and load balancing problem insoftware-defined Mobile Wireless Networks for collaborativecrowdsourcing This paper first presents a network modeltowards traffic engineering problem and then designs a hy-brid routing forwarding scheme as well as a congestion con-trol algorithm to achieve the feasible solution To validate theperformance of the proposals a lot of simulation experimentsare carried out

The rest of this paper is organized as follows Relatedwork in recent years is reviewed in Section 2 The networkmodel is then formulated in Section 3 In Section 4 designof congestion control and traffic scheduling scheme are pre-sented in detail Simulation results and analysis are discussedin Section 5 Finally conclusions are given in Section 6

2 Related Work

At present some researchers have summarized the researchwork of crowdsourcing from different perspectives

Yuen et al in [6] summarized the progress of crowdsourc-ing from applications algorithms performance and datasets Kittur et al in [7] explained the challenges of crowd-sourcing in 12 aspects such as synchronous collaborationreal-time response and dynamic machines Doan et al in[8] reviewed the crowdsourcing system applied on the worldwideweb and summarized the crowdsourcing system accord-ing to the problem type and the way of collaboration Zhaoand Zhu in [9] reviewed crowdsourcing research from fourperspectives information technology the public and orga-nization Kittur et al in [10] have studied how to decomposecomplex tasks and how to integrate workersrsquo answers to per-form initial tasks and proposed a MapReduce framework toachieve the decomposition of tasks However their method isonly suitable for specific types of tasks and the general effectis unsatisfying Scalability still needs to be solved References[11ndash13] focus on the technology of combining machine andhuman with the join operation in crowdsourcing environ-ment which first filters the problem through the machinealgorithm and then assigns the remaining problems to theworkers The authors of [12] used the transitive relationshipsof entities to further reduce the number of tasks therebysaving the cost Lofi et al in [14] reduced the cost of the taskby preprocessing data sets containing missing data throughthe ldquoerror modelrdquo and getting the answers from workersSakamoto et al in [15] studied the ways in which crowdsourc-ing participants often interact in different task types Heeret al in [16] studied how to carry out a survey and found

that the design interface wasmore suitable for crowdsourcingworkers throughquestionnairesThe authors in [17] proposeda method based on random map generation and messagingtask allocation The limitation of this method is that it canonly be used for a specific type of task to the difficulty of thetask However there are various types of task crowdsourcingplatform and some tasks need special professional knowl-edge such as language translation task Liu et al in [18] imple-mented a data analysis system to ensure the quality of theresults as themain goal first through forecastingmodel num-ber assigned tasks and then in the process of task execu-tion through online quality assessment results to determinewhether to terminate the task ahead of time thus saving costand timeThe authors in [19] proposed a new workersrsquo modelin crowdsourcing Through this model the workersrsquo qualitycan be computed accurately and timely For big data tasks thenumber of tasks affects the overall cost of the tasksThe num-ber of tasks can be reduced by effectively designing the taskthus saving the task cost Marcus et al in [20] proposed thestrategy to transform the problem of each task into multiplesubproblems But when a task contains a large number of sub-problems the price of task needs to improve Otherwise itwill be easy to cause only a small number of workers selectedtask That is to say even though such an approach reducesthe number of tasks the overall cost of the task is not guar-anteed to be reduced The authors of [21] presented a com-prehensive system model of Crowdlet that defines the taskworker arrival andworker abilitymodels In [22] the authorsdesigned an approximate task allocation algorithm that isnear optimal with polynomial-time complexity and used it asa building block to construct the whole randomized auc-tion mechanism Compared with deterministic auctionmecha-nisms the proposed randomized auction mechanism in-creases the diversity in contributing users for a given sens-ing job The authors of [23] presented a new participantrecruitment strategy for vehicle-based crowdsourcing Thisstrategy guarantees that the system can perform well usingthe currently recruited participants for a period of time in thefuture The authors in [24] focused on a more realisticscenario where users arrive one by one online in a randomorder The authors in [25] focused on the problem of howto efficiently distribute a crowdsourcing task and recruitparticipants based onD2D communications In [26] existingdefinitions of crowdsourcing were analyzed to extract com-mon elements and to establish the basic characteristics of anycrowdsourcing initiative Based on these existing definitionsan exhaustive and consistent definition for crowdsourcing ispresented and contrasted in eleven cases In [27] the authorsdefined traffic engineering as a large-scale network project tosolve the performance evaluation and network optimizationin the network In [28] traffic engineering has been furtherexplained and the traffic engineering is a route optimizationmethod to improve the quality of network service by avoidingthe link congestion in the network

3 Network Model

There are a number of possible next hops thatmay occur afterthe crowdsourcing task has selected the assignment object in


Next hop 1

Next hop 2

XY loadbandwidth



10100 40100

20100 20100









12 119899



119880 (119890) =sum119890 119871 (119899

119904119889)

119862 (119890) (1)


119904

119889



Minimize 119880 (2)



119875119871(119899119904

119889) ge 0 forall119875 (4)

119871 (119899119904119889) ge 0 forall119899119904119889 isin 119873 (5)



1

2

3

4

5

6Task source node



1

2 3

4 4

5 6 5 6

6 6

Task source node























119889119894119895 = 120590119894 sum119905|(119894119905)isin119864

119888(119894119905)sum119905|(119905119895)isin119864 119888(119905119895)


119894 119895 isin 119881

(6)









Throughput = sum119871 (119899)119879119898

(7)









5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Next hop 1

Next hop 2

XY loadbandwidth



10100 40100

20100 20100









12 119899



119880 (119890) =sum119890 119871 (119899

119904119889)

119862 (119890) (1)


119904

119889



Minimize 119880 (2)



119875119871(119899119904

119889) ge 0 forall119875 (4)

119871 (119899119904119889) ge 0 forall119899119904119889 isin 119873 (5)



1

2

3

4

5

6Task source node



1

2 3

4 4

5 6 5 6

6 6

Task source node























119889119894119895 = 120590119894 sum119905|(119894119905)isin119864

119888(119894119905)sum119905|(119905119895)isin119864 119888(119905119895)


119894 119895 isin 119881

(6)









Throughput = sum119871 (119899)119879119898

(7)









5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



1

2

3

4

5

6Task source node



1

2 3

4 4

5 6 5 6

6 6

Task source node























119889119894119895 = 120590119894 sum119905|(119894119905)isin119864

119888(119894119905)sum119905|(119905119895)isin119864 119888(119905119895)


119894 119895 isin 119881

(6)









Throughput = sum119871 (119899)119879119898

(7)









5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






119889119894119895 = 120590119894 sum119905|(119894119905)isin119864

119888(119894119905)sum119905|(119905119895)isin119864 119888(119905119895)


119894 119895 isin 119881

(6)









Throughput = sum119871 (119899)119879119898

(7)









5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia







Throughput = sum119871 (119899)119879119898

(7)









5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



5

4 3

1

0

6

2

7

(a) 1-SDN node

5

4 3

1

0

6

2

7

(b) 3-SDN nodes

5

4 3

1

0

6

2

7

(c) 6-SDN nodes

5

4 3

1

0

6

2

7

(d) All-SDN nodes


0 1 2

3 4 5

678

9

(a) 1-SDN node

0 1 2

3 4 5

678

9

(b) 3-SDN nodes

0 1 2

3 4 5

678

9

(c) 6-SDN nodes

0 1 2

3 4 5

678

9

(d) All-SDN nodes





0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(c) 6-SDN nodes

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(d) All-SDN nodes




6 Conclusion





Acknowledgments



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



0010203040506070809

1Th

e max

imum

link

util

izat

ion



(a) 1-SDN node

0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n



(b) 3-SDN nodes



0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n

(c) 6-SDN nodes


0010203040506070809

1

The m

axim

um li

nk u

tiliz

atio

n


(d) All-SDN nodes




0

50

100

150

200

250

300

350

Thro

ughp

ut



References











































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia








































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia







RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


Documents

Congestion Control and Traffic Scheduling for ...downloads.hindawi.com/journals/wcmc/2018/9821946.pdf · Congestion Control and Traffic Scheduling for Collaborative Crowdsourcing