Fangzhou Jiang

Demo: Crowd-cache -- popular content for free

Crowd-Cache is a novel crowd-sourced content caching system which provides cheap and convenient content access for mobile users. Our system exploits both transient colocation of devices and the spatial temporal correlation of content popularity, where users in a particular location and at specific times would be likely interested in similar content. We demonstrate the feasibility of Crowd-Cache system through a prototype implementation on Android smartphones.

Type, Talk, or Swype

It is reported that mobile users spend most of their time on texting SMS, Social Networking, Emailing, or sending instant messaging (IM), all of which involve text input. There are three primary text input modalities, soft keyboard (SK), speech to text (STT) and Swype. Each one of them engages a different set of hardware and consequently consumes different amounts of battery energy. Using high-precision power measurement hardware and systematically taking into account the user context, we characterize and compare the energy consumption of these three input modalities. We find that the length of interaction, or the message length, determines the most energy efficient modality. For short interactions, less than 14-30 characters, SK is the most energy efficient. For longer interactions, however, STT significantly outperforms both SK and Swype. When message length distributions of popular text activities are considered, STT provides near optimal energy consumption without requiring the user to predict the message length and decide between SK and STT. In terms of battery life, the choice of input modality makes significant differences. If users always choose SK for all their text activities, they will consume nearly 50% of the phone battery each day. Choosing STT over SK can save 30%-40% of the battery depending on the choice of STT software.

SSIDs in the wild: Extracting semantic information from WiFi SSIDs

WiFi networks are becoming increasingly ubiquitous. In addition to providing network connectivity, WiFi finds applications in areas such as indoor and outdoor localisation, home automation, and physical analytics. In this paper, we explore the semantics of one key attribute of a WiFi network, SSID name. Using a dataset of approximately 120,000 WiFi access points and their corresponding geo-locations, we use a set of similarity metrics to relate SSID names to known business venues such as cafes, theatres, and shopping centres. Such correlations can be exploited by an adversary who has access to smartphone users preferred networks lists to build an accurate profile of the user and thus can be a potential privacy risk to the users.

When to type, talk, or Swype: Characterizing energy consumption of mobile input modalities

Mobile device users use applications that require text input. Today there are three primary text input modalities, soft keyboard (SK), speech to text (STT) and Swype. Each of these input modalities have different energy demands, and as a result, their use will have a significant impact on the battery life of the mobile device. Using high-precision power measurement hardware and systematically taking into account the user context, we characterize and compare the energy consumption of these three text input modalities. We show that the length of interaction determines the most energy efficient modality. If the interactions is short, on average less than 30 characters, using the device SK is the most energy efficient. For longer interactions, the use of a STT applications is more energy efficient. Swype is more energy efficient than STT for very short interactions, less than 5 characters on average, but is never as efficient as SK. This is primarily due to STT enabling the users to complete tasks more quickly than when using SK or Swype. We also show that these results are independent of “user style”, the experience of using different input modalities and device characteristics. Finally we show that STT energy efficiency is dependent on application logic of whether speech samples are for a given period of time before transmitting to a server for analysis as opposed to streaming the speech to a sever for analysis. Based on these observations we recommend that the users should use SK for short interactions of less than 30 characters, and STT for longer interactions. In addition, they should use STT applications which uses storing and transmit logic, if they are willing to trade off battery life to QoE. Finally we proposed the development of an adaptive storing and analyze STT to improve the energy efficiency of it.

uDrop: Pushing drop-box to the edge of mobile network

Content sharing and online collaboration have become extremely popular in recent years, in particular with the pervasiveness of cloud storage services such as Dropbox. However, these popular cloud services are not accessible when users are offline and do not take the geographical location of the users into account when distributing content. Despite years of research in opportunistic content dissemination, real users are yet to fully experience the advantages of content sharing in proximity. To this end, we propose uDrop which literally pushes the cloud storage services to the edge of the network whenever interested users are in proximity of each other. uDrop enables content sharing among users in the vicinity without relying on networking infrastructure while improving the user quality of experience and reducing the cost of bandwidth. In this demonstration, we show the practical feasibility of uDrop service on Android smartphones.