«Opportunistic IoT: Exploring the Harmonious Interaction between Human and the Internet of Things Bin Guo1 Daqing Zhang1,2 Zhu Wang1 Zhiwen Yu1 ...»
The single-broker approach. The crucial issue for SOCKER is to design an appropriate broker-selection approach to facilitate data dissemination. We have proposed the single-broker approach, where brokers are selected based on their social features. Two social features or metrics are used to measure the usefulness of candidate brokers, namely user popularity and user effectiveness. As a basic broker selection metric (i.e., benchmark), user popularity chooses a new broker simply based on the predicted number of contacts the user may encounter in a given period, which is learned from historical contact data. As an improved broker selection metric, user effectiveness additionally leverages the contextual data obtained during the community creation process (using the list of already encountered users to refine user popularity) .
Specifically, each user maintains a list of users that she is likely to meet (learned from user contact history), and the current broker maintains a list of already encountered users (i.e., the context information) since the dissemination process starts. We then calculate the difference-set (DS) of the two user lists, the size of which is used to measure the effectiveness of a user in the broker election process. If the DS of a new encounter is higher than the current broker, broker switch will happen.
Fig. 5: Performance of SOCKER when the community size is set to 5.
We used the MIT Reality Mining (RM)  dataset to evaluate the performance of two different community-creation metrics. The RM dataset contains the co-location information of 106 subjects (staffs and students) at the MIT campus over more than one year. These subjects were equipped with Bluetooth-equipped mobile phones, and their co-location information was collected via frequent (every 5 minutes) Bluetooth device discoveries. To make the dataset more manageable, we have extracted twelve-week of collocation data, corresponding to Sep. 14th to Dec. 7th, 2004. Specifically, the first eight weeks were used as the training dataset while the last four weeks were used as the testing dataset. Meanwhile, while real-world human mobility traces are available, social activity related information (e.g., user preferences/interests) does not exist in the RM dataset. Thereby, we design the experiment as follows: (1) we assume that there are 20 different preferences; (2) each user ui has pi preferences, where pi is a randomlygenerated integer ( 0 pi ≤ 20 ), and each user has five preferences on average; (3) in the experiment, we randomly generate 100 community-creation tasks, where the task initiators and the start time of these tasks are selected randomly.
We tested SOCKER under different community creation expiry time. Figure 5 illustrates the experiment results of the two community-creation methods (i.e., user popularity and popularity+effectiveness). We measure them by calculating the community completion ratio (CCR, i.e., the average ratio of successfully completed tasks) and task transfer cost (TTC, i.e., the average broker-switch times of successfully completed tasks). Both CCR and TTC are calculated within the specified expiry time.
The results indicate that better performance can be achieved when both social features are leveraged in the broker-selection approach.
6.2. Opportunistic Marketing Service When people contact and connect, they influence and exchange the information they own. In opportunistic IoT, peer influence/contact becomes more important than ever, which offers a wealth of new marketing opportunities. For example, we are now developing Opportunistic Trading (as the use case described in the introduction) .
The aim of it is to build a virtual flea market service that works in mobile phone-based opportunistic networks to facilitate request dissemination and match-making among colocated buyers and sellers of goods. An example that illustrates the opportunistic trading process is shown in Fig. 6.
Fig. 6: Opportunistic marketing: an example.
The multi-broker approach. To reduce network cost on data flooding, only sell requests are disseminated, the buyer (while not the seller) is notified when her request is matched. Different from SOCKER, a multi-broker mechanism is proposed, where the buyer can select k brokers and replicate her request to them. At time T5 in the example, S1 and Broker n meets in a coffee shop, and the buyer/seller requests are matched. B1 is then informed of the matched result.
6.3. Community Integration and Orchestration As a promising research direction, we have studied the aggregated effects of heterogeneous community orchestration through two projects: Social Contact Manager and Hybrid Social Networking.
(1) Social Contact Manager: integration of data from heterogeneous networks.
The ability to use the power of a network of social contacts is important to get things done. However, as the number of contacts increases, people often find it difficult to maintain their contact network using human memory alone. People are frequently beset with questions like “Who is that person? I think I met him in Tokyo last year.” Existing contact tools make up for the unreliability of human memory by storing contact information in digital format; however, manually inputting contact data can burden the users. To address this issue, we develop SCM (Social Contact Manager), an intelligent social contact management system . It supports the auto-collection of rich contact data (e.g., profile, face-to-face meeting contexts) from online and opportunistic networks, leveraging the aggregated power of pervasive sensing and Web intelligence techniques.
Fig. 7: Social contact manager: data integration from heterogeneous networks.
Our solution is inspired by the general contact acquaintance process. In social occasions, our connection with a new contact usually starts from exchanging business cards. After obtaining basic information from business cards, people gather more information about the contact from the Web. An interesting phenomenon is revealed, in which the “business card” plays a key role, triggering and leading the contact data gathering process. SCM explores techniques to automate this process, as illustrated in Fig. 7. We employ a mobile card-scanner to extract basic information from the collected business cards (forming an opportunistic network). The scanned basic information is then used to obtain other contact information from the Web (i.e., the online network) using an information extraction method based on a hybrid of heuristic rules and Conditional Random Field (CRF) . The collected information can be leveraged to manage their contacts better, especially for efficient contact retrieval in name-slipping situations .
(2) Hybrid Social Networking: interlinking heterogeneous social networks to facilitate data dissemination. People now connect, interact and transit in heterogeneous social communities (e.g., online, physical, interest/professional groups) within cyberphysical spaces. In the past few years, significant research efforts have been made on facilitating information sharing in online and opportunistic communities. However, they follow separate research lines, and the interlinking of the two forms of communities has little been explored. We have thus proposed the hybrid social networking (HSN) infrastructure , which is inspired by the multi-community involvement and crosscommunity traversing nature of modern people. For example, at one moment, Bob is staying at a place with Internet connection and he can communicate with his online friends (in the online community); later, he may travel by train with merely ad hoc connection with nearby passengers (forming an opportunistic community). Here we use HSN to indicate the smooth switch and collaboration between online and opportunistic communities.
One of the key features enabled by HSN is the popularity-based online broker selection protocol. Different from existing protocols, the online broker selection approach we proposed allows users to choose brokers online from his social connections, while not requiring direct contacts with others in the real world. Users advertise their predicted popularity in the online community, and a publisher can choose the ones with highest-popularity among them. Online broker selection also decreases the time cost on task allocation: the selected nodes can be allocated the dissemination task with no delay if they are online, while offline brokers can be informed of the allocated task once they are within an environment with Internet connection (hotspots, wired network, etc.) We compared the performance of HSN with single-community dependent methods (e.g., the pure ad hoc method), which was also evaluated based on the MIT RM dataset.
We used Opportunistic Trading (depicted in the Introduction) as the background application. As shown in Fig. 6, when using HSN in opportunistic trading, brokers will be chosen from the online connection (or friends) of the buyer. As shown in Fig. 8, for the experiment results of two typical users, great performance improvement (in terms of success rate) is obtained when using HSN. For example, the improvement is about 15% for node 69 and almost 40% for node 79. This is because that the integration of an online community shortens the broker selection process, and increases the opportunity to select brokers with high popularity (in ad hoc or direct contact-based broker selection method, brokers with high popularity may not be encountered and chosen). In summary, the interlinking of distinct social networks can enhance data dissemination among people.
Fig. 8: The effects of hybrid social networking to data dissemination.
7. Conclusion This paper has presented opportunistic IoT, a new research area that addresses information dissemination and sharing within and among opportunistic communities that are formed based on the opportunistic contact nature of human. The bi-directional effects between human behaviors and opportunistic IoT, the co-existing of online and opportunistic communities, as well as the interaction between heterogeneous communities, raise numerous research challenges to opportunistic IoT. Some of them have been discussed in this paper, such as the design of effective protocols on data dissemination considering the impact of human behaviors and mobility patterns, the orchestrating and collaboration of heterogeneous communities in terms of their distinct features, and so on. All these challenges present substantial research opportunities for academic researchers, industrial technologists, and business strategists. We further present four of our ongoing projects/applications on opportunistic IoT, ranging from opportunistic social networking and community service provision, and demonstrate our experience to address the challenges.
In addition to information dissemination, we will explore resource (e.g., built-in sensor resources can be different among users) and service sharing (e.g., different users may keep different services in their device)  within and among opportunistic communities in the future work. We believe that the convergence of anthropology, social science, and pervasive sensing and computing techniques, will greatly propel the development of IoT to its new stage, i.e., stepping into the era of the Social IoT.
Acknowledgement This work was partially supported by the National Basic Research Program of China (No. 2012CB316400), the National Natural Science Foundation of China (No.
61103063, 61222209), the Natural Science Foundation of Shaanxi Province (No.
2012JQ8028), the Basic Research Foundation of Northwestern Polytechnial University (No. JC20110267), the Specialized Research Fund for the Doctoral Program of Higher Education (No. 20126102110043).
References  Zhong N et al. Research Challenges and Perspectives on Wisdom Web of Things (W2T). Journal of Supercomputing 2010;
 Atzoria L et al. The Internet of Things: A Survey. Computer Networks 2010: 54(15): 2787-2805.
 Boldrini C, Conti M, Delmastro F, Passarella A. Context- and social-aware middleware for opportunistic networks. Journal of Network and Computer Applications 2010; 33(5): 525–541.
 Simpson J, Weiner E. Collaboration. Oxford English Dictionary (2nd Edition), Oxford University Press 1989.
 Guo B, Zhang D, Yu Z, Zhou X, Zhou Z. Enhancing Spontaneous Interaction in Opportunistic Mobile Social Networks.
Communications in Mobile Computing (ComC), SpringerOpen 2012 (to appear).
 Weiser M. The Computer for the 21st Century. Scientific American 1991; 265(3): 94-104.
 Hightower J, Borriello G. Location Systems for Ubiquitous Computing. Computer 2001; 34(8): 57–66.
 Gu T, Pung H K, Zhang D. A service‐oriented middleware for building context‐aware services. Journal of Network and Computer Applications 2005; 28(1): 1-18.
 Fischer G. User Modeling in Human Computer Interaction. User Modeling and User-Adapted Interaction 2001; 11: 65-86.
 Campbell A T, et al. The Rise of People-Centric Sensing. IEEE Internet Computing 2008; 12(4): 12-21.
 Blackstock M, Kaviani N, Lea R, Friday A. MAGIC Broker 2: An open and extensible platform for the Internet of Things. In:
Proceedings of IOT’10, 2010.
 Rohokale VM, Prasad NR, Prasad R. Receiver Sensitivity in Opportunistic Cooperative Internet of Things (IoT). In:
Proceedings of ADHOCNETS’10 2010. p. 160-167.
 Wang F, et al. Social Computing: From Social Informatics to Social Intelligence. IEEE Intelligent Systems 2005; 22(2): 79–83.
 Pietiläinen A, et al. MobiClique: middleware for mobile social networking. In: Proceedings of the 2nd ACM Workshop on Online Social Networks, 2009.
 Guo B, Zhang D, Wang Z. Living with Internet of Things: The Emergence of Embedded Intelligence. In: Proceedings of the 2011 IEEE International Conference on Cyber, Physical, and Social Computing 2011. p. 297-304.
 Motani M, Srinivasan V, Nuggehalli P. PeopleNet: engineering a wireless virtual social network. In: Proceedings of MobiCom'05 2005. p. 243-257.
 Karamshuk D, Boldrini C, Conti M, Passarella A. Human Mobility Models for Opportunistic Networks. IEEE Communications Magazine 2012; 49(12): 157-165.
 Mokhtar SB, Mashhadi AJ, Capra L, McNamara L. A Self-Organising Directory and Matching Service for Opportunistic Social Networking. In: Proceedings of the 3rd Workshop on Social Network Systems (SNS) 2010.