ISSN Vol.03,Issue.02, June-2015, Pages:

Transcription

1 ISSN Vol.03,Issue.02, June-2015, Pages: Reliable Sensing for Crowd Management in Cloud-Centric Internet of Things MUDAVATH SOMLA NAIK 1, J. SWAMI NAIK 2 1 PG Scholar, Dept of CS, G. Pulla Reddy Engineering College(Autonomous), Kurnool, AP, India, somla3994@gmail.com. 2 Associate Professor, Dept of CSE, G. Pulla Reddy Engineering College(Autonomous), Kurnool, AP, India, swaminaikcse@gmail.com. Abstract: Modern technology is rapidly altering and enhancing all aspects of our lives. The communication sector has been improving continuously. The Cloud computing is a game changing phase of IT that promises several benefits but the challenges too need to be considered when planning for Cloud adoption. Cloud computing is aimed at providing IT as a service to the cloud users on-demand basis with greater flexibility, scalability, reliability and availability with utility computing model. Cloud computing model can enable provisioning of sensing services through mobile phones, namely Sensing-as a- Service (S 2 aas). In this paper, we study S 2 aas over socialnet working services for crowd management problem where malicious users report false sensor readings leading to severe disinformation at the crowd control platform. To this end, we propose Trustworthy Sensing for Crowd Management (TSCM)which is a reputation-based crowd management scheme over the cloud platform where sensing data is collected from smart phones based on an auction mechanism. Sensing-as-a-Service (S2aaS) is a cloud-inspired service model which enables access to the IoT. The Internet of Things (IoT) paradigm denotes the pervasive and ubiquitous interconnection of billions of embedded devices that can be uniquely identified, localized, and communicated. IoT architecture can be implemented as either Internet centric or object centric. The former aims at provisioning services within the Internet, where data are contributed by the objects whereas the latter aims at provisioning services via network of smart objects. Scalability and cost efficiency of IoT services can be achieved by the integration of cloud-computing into the IoT architecture, i.e., cloud-centric IoT. Keywords: Internet of Things (IoT), Cloud Computing, Crowd Management, Public Safety, Sensing-as-a- Service (S 2 aas), Trustworthy Sensing for Crowd Management (TSCM), Smart Phone Sensing, Social Networking. I. INTRODUCTION Internet of Things (IoT) is increasing the connectedness of things and people on a scale that was not imaginable before. The pace of adoption of IoT in market is increasing due to ubiquitous computing and increase in the connectedness of machines and people. But very few parameters exist to check and ensure the quality of IoT implemented devices and machines. The Internet of Things (IoT) represents a revolution happening now-a-days which results in increased intercommunication and autonomous machine-to-machine data transfer. But with great opportunity comes great responsibility. Along with its conveniences, the IoT unveils several security challenges: in data privacy, safety, governance and trust. Internet of Things (IoT) is technological wave in information technology after the Internet. The IoT connects every physical entity of the world with the internet for switching information and communicate with other entities. It is used to recognize, trace, follow, observe and supervise things. It extends and expands the communication between human and human, or human and machine, or machine and machine where machine can be any physical entity. IoT is achieved through RF identification (RFID) and sensor technology. The Internet of Things (IoT) paradigm stands for virtually interconnected objects that are identifiable and equipped with sensing, computing, and communication capabilities. Implementation of services and applications over the IoT architecture can take benefit of the cloud computing concept. Sensing-as-a-Service (S 2 aas) is a cloud-inspired service model which enables access to the IoT. In this paper, we present a framework where IoT can enhance public safety by crowd management via sensing services that are provided by smart phones equipped with various types of sensors. In order to ensure trustworthiness in the presented framework, we propose a reputation-based S 2 aas scheme, namely, Trustworthy Sensing for Crowd Management (TSCM) for front-end access to the IoT. TSCM collects sensing data based on a cloud model and an auction procedure which selects mobile devices for particular sensing tasks and determines the payments to the users of the mobile devices that provide data. This paper is organized as follows. In Section II, related work and motivation are presented. Section III presents the Cloud based IoT framework, Section IV presents the proposed framework, TSCM in detail. Numerical results are presented and discussed in Section V. Section VI concludes the paper and gives future directions IJIT. All rights reserved.

2 II. RELATED WORK AND MOTIVATION Cloud computing offers on demand access to a shared pool of resources such as software, infrastructure and platform with rapid elasticity based on the pay-as-you-go fashion. Furthermore, cloud computing model harmonizes well with the future Internet of services where several other resources such as CPU, storage and networks will be offered as a service [1]. Provisioning of sensing services through smart phones with in a cloud platform is called Sensing as a Service (S 2 aas). Three main requirements for any S 2 aascloud system as smart phone platform independence, energy efficiency and effective incentives for participatory sensing. In [5], the authors state the advantages of S2aaS as enabling collection, processing and sharing of big data in collaboration of large crowds as a cloud service via mobile devices which already have various built-in sensors. According to [4], introducing S2aaS to social networks would lead to a win-win situation for both systems. Today, social networks connect millions of users based on their interests, relations or features, and it is very likely to evolve to a social network connecting services and applications over the cloud. In [3], based on the requirements of the sensing objects, the authors propose deployment, development, and management of the IoT applications over the cloud, namely, the Cloud Things architecture. Applications that can be improved by the integration of IoT into cloud computing are many; such as pervasive healthcare, smart homes, smart cities, and future transportation systems. Furthermore, public safety in smart city management can be efficiently addressed by taking advantage of cloud and IoT integration. Ina cloud-centric IoT framework, sensors provide their sensed data to a storage cloud as a service, which then undergoes data analytics and data mining tools for information retrieval and knowledge discovery. Built-in sensors in mobile devices can leverage the performance of IoT applications in terms of energy and communication overhead savings. Therefore, Sensing-as-a- Service (S 2 aas) appears as a strong candidate for front-end access to the cloud-centric IoT, where mobile devices provide their sensed data based on the pay-as-you-go fashion. In this paper, we present a cloud-centric IoT-based crowd management scheme for public safety which utilizes S 2 aaswith the ultimate goal of trustworthy crowd sourcing. The proposed scheme is called Trustworthy Sensing for Crowd Management (TSCM) which has been briefly presented in with initial simulation results. As integration of S2aaSand social networks can introduce various benefits to both systems, we present the TSCM framework with the option of social network assistance. TSCM enables the public safety authority to collect sensing data in a particular region for crowd management. The term crowd refers to a large group of people who have gathered for an event. Public safety authority can use sensor data of the smart phones if effective incentives exist for the users in the event to provide their sensing data as a service. TSCM adopts the M Sensing auction-based user centric incentives that have been proposed in for a smart phonebased crowd sourcing scenario, and enhances them by introducing reputation-awareness and trustworthiness of the MUDAVATH SOMLA NAIK, J. SWAMI NAIK smart phone users. Maliciously altered sensing data aiming at disinformation at the public safety authority or inaccurate sensor readings reduce user reputation, whereas truthful and accurate sensor readings increase user trustworthiness.1 TSCM uses recent and past sensor readings of a user in order to compute his/her current trustworthiness dynamically. As TSCM runs an auction to select the users who will crowd source the required tasks, a user s bid and his/her marginal values introduced to the sensing set are redefined by the public safety authority by using the trustworthiness of the corresponding user. We show that when TSCM is adopted, the utility of the public safety authority is significantly improved by reputation-awareness, where as disinformation probability in a crowd sourced task is dramatically reduced when compared to the reputation un aware user-centric incentive approach. Furthermore, public safety authority can save 40% of the payments made to the malicious users in comparison to a reputation unaware auction. III. CLOUD BASED IOT FRAMEWORK The Cloud based IoT Framework removes fundamental constraints concerning place, time and human observations. Cloud centric IoT architecture has following layers in proposed framework, Object Management Cloud Computing Platform Information as a Service Smart Phone User Object Management: Objects have central role in SCM, objects are digital representation of information that communicated and processed via Internet. In SCM objects are information transferred between different partners from primary production to the market. These objects submits sensing task request to the cloud platform and receives sensing data of corresponding task. Cloud Computing Platform: Maintains a user database where objects, sensing tasks and associated events are stored. This middleware enables a seamless exchange of object information between different supply chain participants. Cloud centric IoT framework processes functionalities for service management and service composition. The data exchange and information integration is done by this platform. Information as a Service: This layer considered as data publisher layer in this cloud centric IoT framework for Supply Chain Management. The specific service for different supply chain user based on the object information made accessible by using web interface application. Smart Phone Users: Users at an area of interest are a subset of the crowd to be monitored and controlled by the authority. By installing the S 2 aas application on his/her smart phone, a user agrees to publish the sensing data, as well as his/her bids/payments in any sensing auction over the social media. Furthermore, installation of this application grants access to his/her published S 2 aas data to the TSCM platform. Two types of users are assumed: First type of users bid and publish truthfully whereas the second type of users maliciously end altered sensing data in order to lead to disinformation at the

3 Reliable Sensing for Crowd Management in Cloud-Centric Internet of Things crowd management authority. Furthermore, the second type of users participate the auction with lower bids and aim at winning the auction. IV. TRUSTWORTHY S 2 aas FOR CROWD MANAGEMENT (TSCM) Crowd denotes a group of people who have gathered for an event such as a sports game, a concert or a peaceful demonstration as shown in Fig.1. In all of these activities, the crowd control authorities are responsible for public safety, as well as the safety of the crowd, and they should take immediate action when needed. We propose a reputationaware and social network-based S2aaS framework for crowd management, namely Trustworthy Sensing for Crowd Management (TSCM). TSCM dynamically updates users reputation by using recent and historical data and it jointly considers users bids and their reputation while selecting the participants, as well as making payments to the participants. Fig.1. Illustration of TSCM within the S 2 aas cloud architecture. A. System Architecture The four layers of a cloud-centric IoT, correspond to four main components of TSCM as follows. Crowd Management Authority: Submits sensing task requests to the cloud platform and receives the sensing data of the corresponding tasks once the auction is completed in the cloud. Indeed, sensing task requests also include the value of each task so that the cloud platform can manage the auction by jointly considering the utility of the crowd management authority, as well as the users utility. Cloud Computing Platform: Maintains a user database, where users reputations, bids, payments, sensing tasks, and associated events are stored. Furthermore, the cloud platform interacts with social networking services in order to detect candidate users and collect sensing data. Huge amount of data will need to be stored and processed throughout an event; Hence TSCM over Hadoop Map Reduce framework can be a feasible solution for implementation. Social Networking Service: Is considered to be the sensor data publisher layer in the cloud-centric IoT for crowd management. Furthermore, social network services can cooperate with the cloud platform in retrieving the users who are present at a particular location to attend a particular event. Moreover, S2aaS data such as bids, payments, and sensing data are also published over social network, whereas the above applications enable fast detection of the smart phone users who may take part in the S2aaS auction. Smart Phone Users: Denote a subset of the crowd in a particular region, which is aimed to be monitored by the public safety authority. A user who agrees to publish his/her sensing data on the sensor data publisher layer (e.g., social network) of the IoT architecture, installs the S2aaS application on his/her smart phone. As mentioned in the motivation section, two types of users are considered in this scenario. The first type of users bid and publish their sensor data truthfully (i.e., non malicious users) whereas the second type of users bidlower than their costs in order to guarantee being selected, and publish altered sensing data in order to lead to dis-information at the public safety authority (i.e., malicious users). TSCM consists of the following steps. Step 1: Crowd management authority submits its S2aaSrequest to the cloud platform. Step 2: The cloud platform queries the users who have checked-in nearby corresponding location. Social networks respond with a list of candidate users who will join the auction. Step 3: The cloud platform updates the user database and sends the tasks to the users via corresponding social networks. Step 4: The users publish the sensing tasks in which they are interested on their social networks along with their corresponding bid values, and the social network relays the data to the cloud platform. It is worth while noting that the bid of a truthful user is equal to his/her sensing cost. Step 5: The cloud platform runs the auction. First, it queries the database to retrieve user reputations. Once the user reputations are retrieved, the cloud platform selects the winners of the auction, determines the payments and sends user-task matching information along with the payments to the users via corresponding social network services.

4 MUDAVATH SOMLA NAIK, J. SWAMI NAIK Step 6: Users publish their sensing data on their social TABLE I: Notation used in the Formulation networks, which is also visible to the cloud platform. Step 7: Upon receiving the sensing data, the cloud platform updates its user database with the newly calculated trustworthiness of the selected users. While calculating the trustworthiness of the users, the platform runs an outlier detection algorithm to detect possibly altered data so that trustworthiness of the users who have sent altered data is degraded. Step 8: The cloud platform sends the sensed data of the submitted tasks to the public safety authority. B. Auction Mechanism The auction mechanism in Steps 6 and 7 of TSCM adopts M Sensing auction in and enhances it by introducing reputation-awareness. The auction aims at selecting W winners out of P participants, and determining the payment to be made to each user i (ρi). Table 1 presents the notation used in the model. Among these terms, marginal value of user i for set W (ϑi(w)) is a key parameter of the algorithm, and it denotes the additional value added to the corresponding set by the sensing tasks of user i as shown in (1). Value of a set (ϑ(w)) denotes the total value of the tasks sensed by the users forming the corresponding set as shown in (2),where TW is the list of tasks sensed by the users in set W. Since TSCM introduces reputation-awareness to M Sensing auction, we also use the reputable marginal value (ϑi(w))of user i on set- W as formulated in (3). Reputable marginal value denotes the additional value introduced by user i to the reputable value of set W. Equation (4) formulates the reputable value of set-w which is calculated by summing the values of the tasks forming the sensing set while the value of a task is defined by its actual value scaled by the average trust worthiness of the users participating in the corresponding sensing task (4) We define the trustworthiness of a user during the period t as the ratio of the positive readings (p(t)) to the total readings. When the cloud platform detects a sensor reading as an outlier, the corresponding sensor reading is marked as a negative reading (n).equation (5) is used to calculate the overall trust worthiness of a user, where latest and previous trustworthiness contribute to the overall trustworthiness with β and α coefficients, respectively, where α + β = 1. Thus, the greater the α(β) is, the less the contribution of the latest (previous) reputation is, and vice versa. (1) (2) (3) The auction has two modes, namely, the aggressive and nonaggressive modes. We also define an adaptive mode, where TSCM dynamically switches between these two modes and adjusts α and β parameters. Payment Determination: In the payment determination phase, similar to the benchmark scheme in TSC Mensures that any user, user w, will not be paid less than his/her bid. Initially, each user is assumed to be paid zero. For each selected user, user w, the algorithm constructs a temporary set which is equal to the set of participants excluding user w. TSCM-Adaptive Mode: The adaptive mode of TSCM aims at dynamic adjustment of the impact of recent and past reputation of the user by varying α and β coefficients in trustworthiness calculation. Furthermore, in the winner selection phase (see Algorithm 1),the algorithm first selects the users whose trustworthiness values have not been degraded. This step is followed by selection of the users whose trustworthiness values have been degraded in the most recent task assignment period. To this end, TSC M adaptive mode redefines the reputable marginal value for all users as formulated in (7). Thus, if the trustworthiness of a user has not been degraded in the previous period, the reputable marginal value of the corresponding user for a given set W is set at the naive marginal value of the user for W. By naïve marginal value, we denote the calculation in (2). Otherwise, user s reputation is taken into account in order to calculate his/her reputable marginal value for W, which is defined in (4). (6) V. NUMERICAL RESULTS Performance metrics are defined as follows: Utility of the public safety authority is the difference between the total reputable value of the sensing tasks and the total payments made to the winners in the auction as formulated where τ denotes the τ th period in which the crowd management (5) authority requested a new set of tasks through the cloud platform. Average user utility is the difference between total

5 Reliable Sensing for Crowd Management in Cloud-Centric Internet of Things payments made to the winners and the total sensing cost of the winners. Fig. 2 illustrates the crowd management authority s utility under the reputation-aware framework, TSCM in comparison to reputation-unaware implementation of M Sense. As seen in the figure, reputation-awareness increases the utility of the platform by 12% under lightly arriving sensing task requests and by 85% under heavily arriving sensing task requests when the aggressive mode is on. The reason of this phenomenon is that the users with lower trustworthiness are less likely to be selected in the auction even though their original bids are lower than the other users. Furthermore, the improvement of the aggressive mode of TSCM over non-aggressive mode is at the order of 15%. This is an expected phenomenon since the aggressive mode determines the payments based on the reputable values of the task sets, which leads to cuts in the payments when users with lower reputation (not necessarily to be malicious) participate sensing. Fig.3. Average disinformation ratio. Besides, TSCM is robust to the intermittent disinformation (ID) strategy of the malicious users as the crowd management authority s utility is not significantly impacted when disinformation occurs intermittently rather than occurring continuously. In case of ID, the users improve their trustworthiness by publishing truthful information on their social network. Thus, even though a malicious user is selected, it will most likely publish its sensed data without any alteration. Fig. 4.Average utility of a participant. Fig.2.Utility of the crowd management authority. Fig3 illustrates the disinformation ratio (DIR) as formulated by Eq. 8 under TSCM and the benchmark scheme without reputation-awareness. The improvement under TSCM over the benchmark scheme in terms of DIR is at the order of 75%.Indeed, this is an expected phenomenon as trust worthiness based ranking of the users lower the chance of a malicious user to be selected in the auction. Besides, in case of ID, crowd management authority experiences higher DIR in comparison to the CD case. When a malicious user has the ID strategy, it aims at publishing truthful data until its trust worthiness exceeds the UP THRESHOLD. Once its trust worthiness is high enough to win the auction, it can resume publishing altered data. Therefore, DIR is higher in case of ID. Fig. 4 reports that the improvement in the utility of the public safety authority, savings in the payments to the malicious users, and improvement in DIR are at the expenses of reduction in user utility. However, when compared to the improvement ratio in other performance metrics (Figs. 1 3), the reduction in user utility is significantly low. Furthermore, when sensing task requests arrive more frequently, the degradation in user utility is as low as 2% under the non aggressive mode. Since every selected user is ensured to be paid at least his/her actual bid, slight degradation in user utility is acceptable. In Fig. 5(a), utility of the public safety authority is illustrated for TSCM-nonaggressive mode with ID. Under each load level, introducing reputation-awareness to crowd sourcing increases the platform utility as the ratio of malicious users increases. The enhancement in presence of significant amount of malicious users is more visible under heavy loads, where load denotes the sensing task request arrival rate. Fig. 5(b) illustrates the total payment to the malicious users. In compliance with the previous figure, as the amount of malicious users increases, the benefits of TSCM is more clear as reputation-awareness avoids rapid selection of malicious users in the auction.

6 Fig. 5(c) illustrates the disinformation probability under various malicious user ratios. As users with low reputation are less likely to be selected, disinformation probability can be degraded by reputation-awareness. On the other hand, since reputation-unaware sensing does not discriminate users in the auction with respect to their trustworthiness, as the amount of malicious users increase, it makes the public safety authority become more prone to disinformation. Fig 5: Performance under various malicious user ratios: (a) utility of the safety authority, (b) total payment to malicious users, and (c) DIR. MUDAVATH SOMLA NAIK, J. SWAMI NAIK V. CONCLUSION Today s smart phones offer advanced sensing and navigation capabilities which may promote cloud-inspired provisioning of sensing services, namely Sensing-as-a- Service (S 2 aas). We have proposed a reputation-based framework for a cloud centric IoT architecture with the aim of crowd management, namely, TSCM. In TSCM, the cloud platform confirms the presence of the participants who are willing to collaborate with the public safety authority by publishing their sensing data and bids over their social networks. According to TSCM, the cloud platform confirms the presence of the participants who are willing to collaborate with the crowd management authority by publishing their sensing data and bid over their social networks. Sensing tasks are assigned to the users based on an auction run over a cloud computing platform for TSCM which interacts with the crowd management authority, social networking service and the users. TSCM mainly aims at preventing disinformation caused by maliciously altered sensing data and improving the crowd management authority s utility. We have evaluated TSCM through simulations and shown that utility of the crowd management authority can be improved significantly while disinformation ratio can be degraded by 75%. Another important research finding is that the income of the malicious users who aim at disinformation at the crowd management authority can be degraded by 40%.Early warning systems can utilize S 2 aas to assist disaster management. Malicious groups may also use crowd sourcing to mislead the emergency forces by injecting specious information into crowd sourced data. TSCM can also be adapted to such scenarios to increase community resilience against a crisis. Furthermore, integration of other trust derivation models into TSCM can also enhance its efficiency. VI. REFERENCES [1] BurakKantarci and Hussein T. Mouftah, Trustworthy Sensing for Public Safetyin Cloud-Centric Internet of Things,IEEE Internet of Things Journal, Vol. 1, No. 4, August [2] BurakKantarciand Hussein T.Mouftah, Fellow, Reputation-based Sensing-as-a-Service for Crowd Management Over the Cloud, IEEE ICC Selected Areas in Communications Symposium [3] R. Moreno-Vozmediano, R.S. Montero, and I.M. Llorente, Key challengesin cloud computing: Enabling the future internet of services, IEEE Internet Computing, vol. 17, no. 4, pp , Jul./Aug [4] X. Sheng, X. Xiao, J. Tang, and G. Xue, Sensing as a service: A cloudcomputing system for mobile phone sensing, in IEEE Sensors, Oct.2012, pp [5] S. Distefano, G. Merlino, and A. Puliafito, Sensing and actuation as aservice: A new development for clouds, in 11th IEEE Intl. Symp. OnNetwork Computing and Applications (NCA), 2012, pp [6] X. Sheng, J. Tang, X. Xiao, and G. Xue, Sensing as a service:challenges, solutions and future directions, IEEE Sensors Journal, vol. 13/10, pp , Oct [7] B.B.P. Rao, P. Saluia, N. Sharma, A. Mittal, and S.V. Sharma, Cloudcomputing for Internet of Things and sensing based applications, insixth Intl. Conf. on Sensing Technology (ICST), Dec. 2012, pp

7 Reliable Sensing for Crowd Management in Cloud-Centric Internet of Things [8] W. Tan, M. B. Blake, I. Saleh, and S. Dustdar, Socialnetwork-sourcedbig data analytics, IEEE Internet Computing, vol. 17/5, pp , [9] C. Aggarwal, N. Ashish, and A. Sheth, The Internet of Things: Asurvey from the data-centric perspective, in Managing and MiningSensor Data, C. C. Aggarwal, Ed. New York, NY, USA: Springer, 2013,pp [10] J. Gubbi, R. Buyya, S. Marusic, and M. Palaniswami, Internet of Thing(IoT): A vision, architectural elements, and future directions, FutureGenerat. Comput.Syst., vol. 29, no. 7, pp , Author s Profile:. Mr. M SomlaNaikhas completed B.Tech in Computer Science and Engineering from Y.S.R Engg College of YVU, Proddatur, Kadapa (Dist), AP, in the year Presently he is pursuing him M. Tech from G. Pulla Reddy Engineering College (Autonomous), affiliated to JNTUA, A.P, India. Mr. J Swami Naik, Obtained his B.Tech degree from G. Pulla Reddy Engineering College, Kurnool. He obtained M.Tech from IIT, Guwahati. He is presently working as Associate Professor in the Department of Computer Science and Engineering at G. Pulla Reddy Engineering College (Autonomous), Kurnool, Andhra Pradesh, India.He has got 10 years of teaching experience. His main research interest includes Cloud Computing.