Analysis of server-smartphone application communication patterns

Abstract

The spread of smartphone devices, Internet of Things technologies and the popularity of web-services require real-time and always on applications. The aim of this thesis is to identify a suitable communication technology for server and smartphone communication which fulfills the main requirements for transferring realtime data to the handheld devices. For the analysis I selected 3 popular communication technologies that can be used on mobile devices as well as from commonly used browsers. These are client polling, long polling and HTML5 WebSocket. For the assessment I developed an Android application that receives real-time sensor data from a WildFly application server using the aforementioned technologies. Industry specific requirements were selected in order to verify the usability of this communication forms. The first one covers the message size which is relevant because most smartphone users have limited data plan. The next part discusses reliability issues of the analyzed technologies covering WebSocket connection drop and proxy server caching. Latency tests are conducted as well and the final part discusses the security aspects and how the other requirements are affected when encrypted connections are used. The results show that WebSocket and long polling are relatively good ways to deliver real-time information to smartphone devices. However, the WebSocket connection can drop unexpectedly due to the lack of keep alive messages which are generally not sent on the network. Moreover, latency radically increases when secured WebSocket connection is used.