A SEMINAR REPORT ON
Submitted by:JESNA C S
COMPUTER SCIENCE AND ENGINEERING
SCHOOL OF ENGINEERING
COCHIN UNIVERSITY OF SCIENCE &TECHNOLOGY,
AUGUST 2008 2
Recent years have witnessed the impacts of distributed content sharing (Wikipedia, Blogger), social networks ( Facebook, MySpace), sensor networks, and pervasive computing. We believe that significant more impact is latent in the convergence of these ideas on the mobile phone platform. Phones can be envisioned as people-centric sensors capable of aggregating participatory as well as sensory inputs from local surroundings. The inputs can be visualized in different dimensions, such as space and time. When plugged into the Internet, the collaborative inputs from phones may enable a high resolution view of the world. This paper presents the architecture and implementation of one such system, called Micro-Blog. New kinds of application-driven challenges are identified and addressed in the context of this system. Implemented on Nokia N95 mobile phones, Micro-Blog was distributed to volunteers for real life use. Promising feedback suggests that Micro-Blog can be a deployable tool for sharing, browsing, and querying global information.
Wireless sensor networks export information about the physical environment to computer systems by forming ad-hoc topologies and aggregating or streaming thousands of local observations to a central location . This model has been most successful in scientific and military settings such as wildlife surveillance, volcano monitoring , vehicle tracking , and environment sensing . These networks are practical because sensor nodes such as Cross- bow motes provide an attractive price-performance ratio . Motesâ„¢ low cost makes install- ations of hundreds or thousands of observation points feasible. The affordability of such large networks allows deployments to cover large geographic areas and creates redund- ancy for overcoming the high failure rate of individual nodes. Yet despite being relatively inexpensive, motesâ„¢ computational and communication technologies are powerful enough to support a wide range of complex distributed applications.Mobile computing hardware is approaching a similar priceperformance moment in which powerful, low-cost commodity devices will enable a new generation of participatory sensing applications . In particular, commodity mobile phones with a variety of sensors (cameras, GPS, accelero- meters, healthmonitors) are already widely deployed and becoming more ubiquitous each year. Each phone can be viewed as a virtual lens, capable of focusing on the context surrounding it. By combining the lenses from billions of active phones, it may be feasible to build a virtual information telescope that enables a high-resolution view of the world. This vision is illustrated in Figure 1 â€œ Micro-Blog aims to translate this vision into a widely deployable service. The central idea is as follows: Mobile phone users are en- couraged to recordmultimedia blogs on-the-fly, enriched with inputs from other physical sensors.The blogs are geotagged and uploaded to a remote server that positions these blogs on a spatial platform (e.g., Internetmaps). Internet users can zoom into any part of the map and browse multimedia blogs at those locations. Moreover, users may query se- lected regions for desired information. Queries are serviced either through explicit human participation, or automatic physical sensing.One of the key technological challenges faced by Micro-Blog is that the application needs to resolve a tension between localiza- tion accuracy and battery lifetime. Devices often have access to multiple localization ser- vices, and reasoning about each of their tradeoffs is non-trivial. For example, GPS can be
accurate to within several meters, but can drain a mobile phoneâ„¢s battery within seven hours. WiFi based localization can improve battery lifetime at the expense of higher loc- alization error, and is limited to dense war-driven urban areas. GSM localization is widely available and extremely energy efficient, but is accurate to hundreds of meters. Micro-Blog addresses this challenge by switching between different schemes with the aim to achieve an application-specific balance between accuracy and battery life.
32. MICRO-BLOG SERVICE DESCRIPTION AND APPLICATIONS
User generated/sensed content are posted on a location-based map. Internet users browse and query this content. Queries are resolved through participatory or auto- matic sensing. Figure 1 illustrates the technological overview of Micro-Blog. Taking advantage of phone sensors, users are encouraged to record multimedia blogs on-the-fly. Blogs may be enriched with a variety of sensor inputs, such as accelerometer vibrations, health sensors, WiFi SSIDs. The phone application associates the blog with the time and location of the device, creating what we call a microblog. The microblog is transported over an avail- able wireless network such as WiFi or cellular, to a web-accessible database. The microb- logs are then positioned on a spatial platform, such as Google Maps or Microsoft SensorMap .A variety of web services can be used to mine, group, and correlate blogs based on interests, themes, and social networks. An Internet user can zoom into any part of this world map, and browse streaming content originating from that region.
Querying in Micro-Blog : When specific content is not available on the map, users can mark out a geographic location and direct queries to phones located near it. Human re- sponses to these queries can be placed on maps as well, facilitating a platform for loca- tion aware content sharing over mobile devices. The responses may also be stored in a database, and later used to answer new queries that seek similar information. Since the queryresponse interaction occurs in human languages, it can be sophisticated and often subjective in nature. Issues such as incentives and privacy are also pertinent challenges for the micro-blog platform. Importantly, not all queries may warrant explicit human par- ticipationi) One category of queries may require the phone to automatically sense and dispatch events. Accelerometer readings from phones of automobile riders , or open WiFi SSIDs near a street cafe are two examples. (ii) A slightly different category may require a very low level of individual participation. In this category, a query might have a long ex- piry time, and may suffice if at least one among many people replied to it over a large time window. The quality of tourist experience at an island could be one example. Micro- Blog enables the entire range of location-aware querying, without any guarantees on whether the queries will be serviced. Live case studies with Micro-Blog suggest that users are cognizant of the lack of guarantees and are yet interested in asking questions to different parts of the world.
Micro-Blogs at Physical Locations: In addition to superimposing them on virtual space (Internet maps), microblogs may also be superimposed on physical space. An individu- alâ„¢s microblog about a restaurant may float near the restaurant, and can be pushed to other peopleâ„¢s phones when they arrive at that physical location. A general publish/sub- scribe system can evolve that automatically sifts through floating microblogs, and draws user-attention to those of interest. Further, the database of floating microblogs can be queried, modified, or updated through the interface of mobile phones and social participa- tion.
In this section, we describe several potential applications that may emerge on the Micro- Blog platform. 2.1.1Tourism
: Imagine an internet map (e.g., Google Maps) punctuated with numerous icons, each having a Play button. An Internet user, Jack, planning a vacation to a beach, navigates to Carolina Beach, USA, and clicks on one of the Play buttons. An audio- visual microblog is played that describes an anonymous touristâ„¢s experience at the beach while she was there a few weeks back. Although useful, Jack decides to browse a few other blogs located around the same geographical region. Unfortunately, none of the blogs he played discussed parking facilities near the beach, an important criterion in Jack- â„¢s beach selection. Therefore, Jack records his query on his laptopâ„¢s microphone, marks out a destination region around the beach, and sends the query to phones in that region. Since he wants to finish planning soon, he specifies the query to be valid for 1 hour.Tour- ists located within the marked region of the beach, those that have Micro-Blog services turned on, find a pop up query on their mobile phones. Only a few people respond by re- cording their replies into their phones and adding pictures of the parking spots (a new mi- croblog). Soon, Jack sees the visual details about parking at the Carolina Beach and is happy with the situation. While nearly convinced, he directs just one more query to phones on the beach, requesting information about availability of WiFi access points. WiFi interfaces on some phones resolve this query and return the scanned SSIDs and sig- nal strengths â€œ the phone user does not participate in generating this reply. Satisfied with the outcome of his research, Jack proceeds to search out a good hotel using Micro-Blog.
: It might be feasible to develop a news service in which citizens play the role of journalists, reporting audiovisual events in the context of their lives. News may also be derived from automatically sensed information from mobile phones, such as the air quality of a region of interest. Subscribers to such news channels can search extremely high resolution news about any place/topic/event of interest. The spon- taneity of capturing information with phones, and the ease of publishing them, can make such a news service different from Internet based blogging sites.
: Superimposing microblogs on physical space can act as location- aware alerts. When walking paths in a university campus are not available on Internet maps, microblogs can be virtually placed at different spots with instructions to reach dif- ferent department buildings. A lost user, trying to reach Building X in the university, can download pre-recorded walking directions from her current location to building X. Meta- phorically, micro-alerts are like virtual sticky notes floating on physical space.
: In general, applications on Micro-Blog may connect those who need services to those who can offer them. (1) We envision the possibility of on-the- fly car pooling where ride-takers can dynamically establish a connection to nearby ride- givers, reducing fuel consumption. (2) Rural healthcare may benefit from Micro-Blog. Surveys have shown that villagers have a high financial/logistic cost of visiting doctors, and often underplay their illnesses as an excuse to postpone medical visits. Even slight indications about the potential gravity of a disease can be effective in arresting a train of mistakes. Micro-Blog can be a platform for providing low-cost, preliminary, health-indic- ations. Medical practitioners can browse microblogs posted from villagers of the world, and offer preliminary indications (e.g. you need to see a doctor immediately). In con- trast to unavailable Internet access, the widespread adoption of mobile phones may be particularly amenable to Micro-Blog. (3) Similar applications may also be feasible in the context of education, where students from rural regions post their queries, and one among many qualified users, respond.
3. ARCHITECTURE AND DESIGN
Figure 2 presents the client-server architecture of Micro-Blog. Recorded blogs are labeled with the locations and access permissions. A localization service is consulted to obtain a reasonable location estimation. As mentioned earlier, even when GPS is available, its continuous use may not be energy-efficient. Hence, the localization service selects one among several localization schemes (WiFi, GSM, GPS, or combinations thereof) that meet the applicationâ„¢s accuracy/energy requirements. The location-tagged blog is trans- ported to an application server over a WiFi or cellular connection, which in turn forwards it to a back-end database. The blogs are organized/indexed appropriately, based on blog originators, access permissions, themes, or other keys. When a client wants to access a microblog, it contacts a web server to retrieve it. The web server provides only those blogs that the user has permissions to access. Phones periodically update the Micro-Blog
server with their own locations, which maintains per-phone location in its localization database. When a user sends a query to a specific region, R, the server first determines if the query can be serviced from the database of blogs. If feasible, the server retrieves all microblogs that match the time, location, and permission attributes, and returns them to the user in reverse chronological order. Otherwise, the server selects phones located in the region R (that have declared themselves available) and forwards the query to them. Phones that arrive later at R also receive the query. When some phone responds to this query, the response is linked to the query and placed on the map as a new, interactive mi- croblog. Queries are active for a pre-specified lifetime, configured by its originator. Upon expiry, they are removed from the server. To achieve content distribution in physical space, the Micro-Blog server also pushes" location-specific blogs to phones that arrive at that location. The push operation is triggered when the phone updates the server with its location. Of course, not all microblogs are pushed to phones â€œ only channels that have been subscribed to by the respective user.
3.1 Design Considerations
Several technological challenges and social concerns underlie mobile phone based ap- plications. Examples include (1) exhausted battery in times of need, (2) efficient localiza- tion accuracy, (3) incentives for participation, (4) privacy, (5) spam, and several others. Some of these issues exhibit inherent tradeoffs and must be addressed in a manner that suits the application. This section presents a pertinent range of design issues and tradeoffs and discusses proposed approaches applicable in Micro-Blog.
3.1.1 Energy-Aware Localization
Issue: Location information is central to Micro-Blog. Not only is content tagged with locations, phones too must periodically report their own locations to the remote server such that they can be suitably queried. While GPS based localization provides good ac- curacy (around 8 meters), measurements demonstrate an unacceptable battery life of less than 7 hours, when GPS is used continuously. Even if Assisted-GPS (AGPS) is used, the improvement is only marginal. Clearly, energy-aware localization services need to be de- signed to balance the energy and accuracy demands of practical mobile phone applica- tions. Proposed Approach: Alternate localization schemes have been well studied and are widely available. Research from Place Lab has proposed a variety of solutions using WiFi based or GSM cell tower based localization. However, energy-accuracy tradeoffs apply to these techniques as well. Energy consumption with WiFi based localization, al- though better than GPS, is significantly worse than GSM schemes. A judicious selection needs to be performed based on the application needs, residual battery power, and a phoneâ„¢s mobility characteristics. Instead of choosing a static localization scheme, it might be beneficial to switch between different schemes to achieve a better energy-accuracy tradeoff. New schemes may also be designed that offer new operating points in this tradeoff space. For instance, if collaboration between mobile phones can be incentivized, it may be possible to opportunistically localize one phone using GPS beacons from oth- ers. Schemes may further adapt to residual battery power by switching to aggressive loc- alization when the phone is predicted to get recharged soon, and vice versa. While a host of possibilities exist, we develop a simple energy-aware extension of Place Lab to suit Micro-Blog applications. In Place Lab, authors create a wireless map of a region by war-driving in the area. The wireless map is composed of sampled GPS locations, WiFi access points audible at these locations, and their corresponding signal strengths. This wireless map is then distributed to phones. When a phone travels through the mapped area, it estimates its own location by matching its list of audible WiFi APs to the wireless map. Several matching schemes have been proposed, including (1)Centroid Computation, where the centroid of all over- heard APs is assumed to be the position of the phone, (2) Signal Strength based estima- tion, where signal strengths are used as weights in computing the weighted average of the audible APs, and (3) Fingerprinting, where the location of the phone is assigned to a point in the wireless map which is closest in terms of an euclidian metric. Experiments show that Place Lab is capable of achieving a median positioning error of 13 to 40m. Employing a similar technique with GSM cell towers increases the localization error range from 94 to 196m, but provides a greater service area. The experiments were mostly per- formed in the downtown area of Seattle with reasonable WiFi and GSM coverage. Micro-Blog augments Place Lab with an energy-aware optimization. The basic idea is as follows. A phone switches between multiple localization schemes, such that the localiza- tion accuracy can benefit from the more accurate scheme, while battery life can improve due to themore energy-efficient scheme. Without loss of generality, we explain the strategy using WiFi and GPS only.Using our scheme, a phone remains in the WiFi mode by default,and periodically computes its estimated location. When it finds that the WiFi fingerprint is not changing over time (an indicator of no macro movement), it determines its GPS location once, and uses it for subsequent blogging functions. The location error is expected to remain minimal while the node is stationary1. However, once the phone de- tects movement(inferred from changes in audible WiFi SSIDs) it begins estimating the distance it has moved and the approximate velocity. When the distance moved becomes greater than a particular threshold, !, the phone records its GPS location once again, and computes the current error in WiFi localization. Let t0 denote any time instant at which the phone takes a GPS reading. In subsequent time steps, ti, the phone estimates its loca- tion through a simple linear interpolation using the precise GPS location from the past, Lg(t0), and its WiFi-estimated location, Lw(ti). The weights of the linear interpolation is derived from the estimated velocity, vest. More precisely, at time ti, the interpolated loca- tion is computed as Lw+g(ti) = Lg(t0) + i Ãƒâ€” vest along the straight line joining Lg(t0) and Lw(ti). The interpolation is performed for " time steps, after which the phone does not utilize the knowledge of its past GPS measurement. When the phone has moved greater than ! distance since time t0, it samples its GPS location again. Figure 5 shows the pro- posed interpolation scheme. The same idea can be applied for switching between GSM and WiFi. Each of these combinations will lead to a new operating point in the energy- localization tradeoff space, offering the application to select one that suits it best. Of course, this simple interpolation scheme does not achieve optimality by any means. Moreover, if users take sharp turns in their movement path, it may not trigger a change in the WiFi-estimated location for a long duration. As a result, the proposed scheme will continue to extrapolate the original direction of movement, potentially resulting in greater inaccuracy. The correction will occur only when the WiFi fingerprint is itself updated (i.e., Lw(ti) changes). Furthermore, the parameters of ! and " need to be carefully chosen, based on mobility patterns, desired localization accuracy, and energy budget.We believe energy- aware localization in mobile phones is an open research problem, and merits individual research attention. Developing optimal, application-aware solutions is a part of our ongo- ing research.
3.1.2 Energy-Aware Applications
Issue: Energy-aware decisions are often influenced by the application requirements. From the end-user perspective, some operations may be sacrificed for improved energy-effi- ciency, while others may be critical for acceptable performance. Proposed Approach: We have incorporated two simple forms of energy awareness in Micro-Blog. (1)When a blog is created by the user, the application records the current GPS location to maximize the accuracy of blog placement (on Google map). However, when a phone periodically reports its location to the Micro-Blog server, it uses a WiFi or cell-based loc- alization, or combinations thereof. Since the exact location of the phone may not be crit- ical for spatial querying, lower accuracy is tolerable.(2) Micro-Blog delays uploading blogs if it finds that the residual battery power is below some threshold. The blog is trans- ported only after the phone has gained sufficient charge, such that the residual energy does not go below the threshold even after expending the blog-transmission energy. The default threshold is set to 72% of the battery life, and is a parameter that can be changed by the user. The transmission energy is calculated as a function of the blog-size. While postponing blog transmission increases the delay of posting blogs, it assures users that the phone retains energy for unforeseen emergencies. User feedback show that the facility to set this threshold was a valued feature.
Issue: As described, querying Micro-Blog assumes that users respond to queries even when they are generated by strangers. A natural question to ask is why would anyone agree to reply to a strangerâ„¢s queries when replying costs battery power and the userâ„¢s at- tention? Proposed Approach: We propose to handle incentives through two mechanisms: social networks and explicit incentive mechanisms.(1) Queries could be restricted to within so- cial networks. In such cases, users may not need a well defined incentive to reply to quer- ies from friends. While this approach facilitates participation within a social group, it does not encourage communication between strangers, confining content distribution only within social clusters.(2) In the absence of an underlying social network, Micro- Blog can apply a give and take approach to incentives. Initially,a user would be re- gistered with a pool of n free query credits. Every time she initiated a new query, her count would be decremented, while every response she generated would increase her count by K (K " 1). The value of K can vary based on the global balance of queries and responses. In the worst case, K could be 1, requiring each user to respond to one query before initiating a new query. However, if the system exhibits healthy participation, K could be increased. Unfortunately, colluders could also initiate bogus queries and re- sponses to artificially inflate their query totals. We draw upon existing graph theoretic ap- proaches to address this problem. A directed link between two nodes, (i, j), can denote a query; the cost of the link can represent the number of queries from i to j. If a clique of nodes are identified such that the cost over all the links are greater than a threshold, then these users will be under suspicion of misbehavior. These users can then be penalized either through a reduction in their query credit, or by some other mechanism. Incentives are also necessary to cope with the financial costs of using Micro-Blogs. Mobile phone service providers typically charge for data transmission/reception, and microblogging multimedia content can incur non-marginal costs. However, with commoditization of mo- bile phones, applications and services will play a critical role in customer retention. Trends show that services will be offered free to customers, as a value added package to basic voice communication. Micro-Blog is envisaged to be one of the applications in- cluded in such a package.
3.1.4 Location Privacy
Issue: Placing blogs at its location of creation reveals the where abouts of the blogâ„¢s ori- ginator. This presents several location-privacy challenges. Proposed Approach: Since users periodically transmit their location to a central server, they must trust server administrators to manage this data properly. This appears to be a fundamental drawback of the Micro-Blog architecture, although techniques from related work that allow clients to securely exchange locations through a mutually untrusted inter- mediary may be applicable . Nonetheless, we currently assume that users are willing to trust the Micro-Blog service. In the simplest case, a Micro-Blog client operates in three modes: private, social, and public. The private mode can be restricted to individual view- ing only â€œ e.g., microblogs about a userâ„¢s honeymoon trip may be tagged private, mean- ing that it can only be viewed by the user. The social mode could only be visible to mem- bers in the userâ„¢s social network, while the public mode can be open to all Micro-Blog users. A userâ„¢s current location can be tagged similarly. A user that wishes to be social may be forwarded queries from members in her social networks. If the user is willing to be public, anyone may send her a query. When in the public mode, a user may choose to assume a pseudoidentity such as a nickname. In addition, a user can obfuscate her loc- ation through her localization service. For example, switching a device into cell tower localization intentionally introduces uncertainty about a userâ„¢s whereabouts. Micro-Blog provides a GUI on the mobile phone that allows users to control these privacy prefer- ences.
Issue: The problem of unwanted queries or spam is similar to the incentives problem de- scribed in Section 4.1.3. We apply a similar solution. Proposed Approach: We allow users to regulate the type of query traffic that they are willing to receive. While a user is in social mode, she may wish to receive at most Qss queries per day from members of her social network only. Of course, Qss can be zero. In the public mode, she may wish to receive at most Qps queries from her social network, and at most Qpp queries from the public. We set the default values as: Qss = 5, Qps = 5, and Qpp = 2. The values can be configured at any time. When a user activates Micro- Blog or changes these configurations, the phone updates the server with these parameters. The server blocks queries in accordance.
3.1.6 Content Inaccuracy
Issue: Because microblogs can be generated without any centralized oversight, their content can reflect (intentionally or not) inaccurate information. Proposed Approach: In public mode, users can be assigned reputations, based on the ratings other browsers assign to her microblogs. If a userâ„¢s microblogs continue to receive low ratings from others, then her reputation will decrease. The reputation of the user is translated into a confidence value reflecting the serverâ„¢s confidence in that blog. When a user is below a threshold reputation, microblogs from that user is not published. While this scheme cannot immediately identify a false content, over time, a misbehaving user can be identified.Of course, Micro-Blog has no way of preventing users from ex- ploiting a strong reputation to make an inaccurate microblog more believable.
4. SYSTEM IMPLEMENTATION
The Micro-Blog phone client, implemented on Nokia N95, is approximately 6,000 lines of code using the Carbide C++ compiler supported by the Symbian OS platform2. The server is approximately 7,500 lines of code across C++, PHP, Ajax, and MySQL.
4.1 Phone Client
The Micro-Blog phone client has been developed using Nokiaâ„¢s Carbide C++ (version 1.2) in theWindows environment. The software requires access to the location services, network services,and the file-system on the phone. The Nokia S60 3 rd edition SDK provides the necessary APIs for this. However, Symbian applications need to be signed with appropriate certificates.We obtained a developers certificate based on thephoneâ„¢s IMEI identifier.The Micro-Blog client is designed to minimally interrupt thenormal oper- ations of the phone. Figure 6 presents the state diagram for the software. By default, the client is in the Idle state. When the user initiates a new blog, the client makes a transition to the Blog state and tracks all the user-generated content being added to the file system. Once completed, the content is tagged with the phone location. To allow offline blogging, e.g., blogging about a vacation while driving back home, Micro-Blog allows users to ex- plicitly modify the location of a blog. The user can visit a map on the phone, determine the location of where the pictures/videos were taken, and assign these coordinates to the microblog. Once the user clicks the send button, the client switches to the Sync stateand transports the blog to the server over a TCP connection. The software then returns to the Idle state.Micro-Blog implements a periodic polling mechanism to monitor inputs from phone sensors. The application transfers to the Log state whenever a poll returns a new result, or if some of the asynchronous sensors trigger an alert. Sensor data,including loca- tion, WiFi SSIDs, GSM tower IDs, and signal strengths, are logged on the phoneâ„¢s file- system. Upon completion,the client returns to the Idle state.When a user configures the phone to be visible on the map,the client application initiates and maintains a persistent TCP session with the remote server. Location coordinates are periodically uploaded to the server, along with requests for new queries. If the phone qualifies for some pending queries (depending on permissions, social memberships, etc.), the queries are forwarded to the phone. The client transitions to the Query state, displays the query on the phone, and waits for user action. Once the reply is created, the phone transports it back to the server and returns to the idle state.
4.2 Web Infrastructure
5. RELATED WORK
Recently, a new wave of research is focusing on developing participatory platforms for people-centric mobile computing applications. We present some of the related work in this road direction. Project Satire aims to develop smart clothes that will monitor human activity and trigger alerts in specific situations. Using accelerometer-equipped sensor motes, authors have shown the possibility to discriminate between human actions based on their accelerometer signatures. Project BikeNet has demonstrated an application in which biking experiences of humans can be monitored and shared within a social circle. Through a variety of sensors attached to bikes, bikers are characterizing the experiences of bike riding over a given path. Several other applications closely relate to human activ- ity, such as PeopleNet , MAX , SenseWeb , My-Experience , MetroSense More recently, the widespread adoption of mobile phones, and the tremendous growth in their comput- ing/communication power, have made them an attractive platform for sophisticated ap- plications. Several systems have begun to emerge and authors have demonstrated the pos- sibility of estimating traffic congestion using accelerometer readings and GPS traces from mobile phones. A range of projects are attempting to aggregate on-body health data into mobile phones with the end goal of building a health monitoring and emergency re- sponse system. Project MetroSense is developing a generic platform for urban sensing through a combination of mobile phones and sensor motes. In the population of mobile phones, combined with user inputs, have been used to dynamically estimate the demo- graphics of a region. Authors describe applications in which the visitor frequency of a tourist spot, or the spatial contour of certain gasses, can be visualized by combining the inputs from multiple phones. Several other systems and platforms are assembling mobile phone based systems that will enhance our view of of the physical/biological world we live in. While many systems are oriented toward sensing the physical/chemical/biological surrounding, a concurrent trend in social computing and networking is becoming dis- tinctly visible. Recently referred to as Web 2.0, this trend is exhibiting the power of com- bining human participation with mobile technology . Authors in have revealed the possibility of video communication as one way of sensing the urban environment. They suggest the possibility of performing streamed video blogging, directly from public mobile phones to content subscribers on the Internet. The authors study the prospects of using mobile computing to facilitate participation and social adoption of networking techno- logy. The study highlights interesting insights on how collaboratively participating in technology impacts the social relationship and community formation in developing re- gions of the world. Nokia has launched Lifeblog that allow users to create a timeline of their activities (photos, sms, video, etc.) â€œ a type of digital recording of oneâ„¢s own life. Web services such as Twitter , Dodgeball , etc. are enabling people to explicitly commu- nicate their status to others within social networks. A slightly different type of service enables users to find other people of similar likings around them, and invite them to meet up on physical or virtual space. When injected with location awareness, many of these applications are becoming even more interesting. Project AURA demonstrates a proto- type in which mobile phones read bar-codes from on-shelf products, and query Internet search engines to obtain product ratings and information. On similar lines, project Cooltown summarizes its objective as allowing everything to have a web . In another interesting project , messages are pushed to phones once they arrive at a location. For ex- ample, Socialight allows users to go to different places in the downtown New York, and receive memos about neighborhood activities. Authors in propose a similar location- aware reminder system â€œ users schedule reminders on their phones that are triggered only when they reach a location. Micro-Blog is a generalization of the above ideas, augmented with the ability to query mobile phones, or their human owners. Responses to these quer- ies can contain a mix of quantitative and qualitative information, often useful for decision making. Moreover, human participation may facilitate distillation of raw sensor data by combining it with the context in which they are generated. The framework to exploit hu- man processing as an instrument to overcome the limitations of device-based sensing, can potentially lead to a new paradigm in information technology. Micro-Blog attempts to facilitate such a paradigm.
6. DEMO AND ONGOINGWORK
Figure 1 shows the simplified architecture for Micro-Blog. We have implemented a light weight Java client on the Mobile phone, a bluetooth/WiFi/cellular based wireless routing protocol, a Micro-Blog web service, and a visualization front end. Our system enables a phone to transport a microblog either over a single hop connection to the server, or over multihop connections as in delay tolerant networks (DTN) . Simple querying is alsofeas- ible, as long as the phone has an established session with the Micro-Blog server. Our web service and visualization front end receives the microblog, processes them, and places them on Google maps as shown in Figure 2. A fully functional system entails a widevari- ety of challenges spanning over a variety of research areas including wireless networking, mobile computing, databases, data mining, and economics.In this demo, visitors will be given GPS-enabled mobilephones, and asked to blog. The phone will automatically create the microblog using the recorded media contents,GPS, and time, and opportnistically route it (overmulti-hopif necessary) to the Micro-Blog server. When the bloggercomes back to our demo site, the microblog should to be should be posted at the appropriate loc- ation. Querying will also be demonstrated, however, since we have very few GPS en- abled phones, the demonstration will only be a proof-of-concept. Ongoing work is directed towards addressing a number of research problems. (1) An incentive sche- meneeds to be designed. (2) An opportunistic routing protocol is necessary to route mi- croblogs from phones to the nearest internet gateway (can also be a cell tower). Also, a location-aware broadcast mechanism is necessary for propagating queries.(3) The sys- temneeds to be energy-aware in view of the limited battery on the cell phone. (4) The web service will receive streaming blogs and queries, that must be processed/ grouped in a variety of user-specified ways. Several database related challenges underlie the scalabil- ity of such a system . In the future Micro-Blog may converge with passive sensor net- works that sense events through medical devices, accelerometers, video cameras, etc. The ongoing work is systematically addressing these issues with the goal of translating this prototype into a human-usable system. drawn box around a Chinese temple. Phones within the box respond as shown in the left panel. 7. LIMITATIONS
Mobile phone based participatory sensing is an emerging platform; several issues need to be addressed across multiple disciplines. This section discusses several limitations with Micro-Blog, with some references to our ongoing work. .
7.1Privacy in General Public
: Our user study was performed with university students who may not be representative of the broader population. General users may not be will- ing to expose their real-time locations to a global forum. Schemes need to be developed that effectively obfuscate individual user locations, while maintaining the population stat- istics. Hence , one should be able to know the number of people in Manhattan, without knowing any individualâ„¢s location. Even the service provider should not be able to locate an individual. Micro-Blog currently does not support such standards of location-privacy. 7.2Exploiting Accelerometers for Localization : Our multi-modal localization scheme exploits a userâ„¢s movement patterns to address the energy-localization tradeoff. For instance, in the absence of macro movements for a threshold time, the phone invokes a GPS measurement. In this paper, user movements have been estimated through changes in WiFi or GSM based locations. These are low-accuracy indicators. Utilizing accelero- meter readings from a phone, as well as an userâ„¢s mobility profile, can lead to improve- ments. We are designing an energy-aware localization scheme that takes advantage of both these opportunities. Kalman filtering techniques may be particularly applicable in this context.
7.3Integration with a Social Network Application
: Several attributes of Micro- Blog rely on a pre-established social network. It may not be practical to initiate a new so- cial network for Micro-Blog users. Integrating Micro-Blog with Facebook or similar ser- vices is perhaps a better option. We are making progress towards this direction.
7.4Identifying False Content
: Reputation based schemes have been used in the past to identify false information. However, such schemes have a high latency to converge on a misbehaving user. An intelligent attacker can always operate below the threshold of de- tection, and may be successful in misleading others with false data.Since Micro-Blog may be employed for real-time decision making (e.g., whether to take a side-street to avoid traffic congestion), potential security concerns may need to be addressed.
In this paper, we presented the design and implementation of a participatory sensing ap- plication called Micro-Blog, which allows smartphone-equipped users to generate and share geotagged multimedia called microblogs. This data can be browsed or queried through either an Internet map service such as Google Maps or in physical space as a user moves through a location. In addition, Micro-Blog allows an Internet user to directly send queries to a set of phones located in a region of interest. Phone users in the specified re- gion that receive these queries can then reply at their discretion. One of the key technical challenges addressed by this work is the need to balance the competing goals of accurate location coordinates and long battery life. Because microblogs are geotagged, user exper- iences depend on accurate matches between data and a location. On the other hand, ac- curate localization cannot come at the cost of unacceptably short battery lifetimes. Micro- Blog addresses this issue by infrequently using more accurate, but power-hungry localiz- ation services such as WiFi to offset the error introduced by less accurate, but more power-efficient localization services such as GSM localization. We evaluated Micro-Blog by building and deploying a prototype implementation. Preliminary experience with users has been highly promising; while challenges such as nonacademic user psychology, in- centives, and privacy remain unmet, usersâ„¢ overall experience with the system was posit- ive.As a result of this feedback and our energy-efficient localization scheme, we believe that Micro-Blog represents a promising new model for mobile social collaboration.
1.R. Szewczyk, E. Osterweil, et al., Habitat monitoring with sensor networks, Commun. ACM, vol. 47
2. S. Gaonkar and R. R. Choudhury, Micro-blog: Mapcasting from mobile phones to virtual sensor maps,