Organizations are starting to realize the significant value of advertising on mobile devices, and a number of systems have been developed to exploit this opportunity. From a privacy perspective, practically all systems developed so far are based either on a trusted third-party model or on a generalized architecture. We propose a system for delivering context, location, time, and preference-aware advertisements to mobiles with a novel architecture to preserve privacy. The main adversary in our model is the server distributing the ads, which is trying to identify users and track them, and to a lesser extent, other peers in the wireless network. When a node is interested in an ad, it forms a group of nearby nodes seeking ads and willing to cooperate to achieve privacy. Peers combine their interests using a shuffling mechanism in an ad-hoc network and send them through a primary peer to the ad-server. In this way, preferences are masqueraded to request custom ads, which are then distributed by the primary peer. Another mechanism is proposed to implement the billing process without disclosing user identities.

Typically, oracles used to test graphical user interface(GUI) programs highly depend on environmental factors that are not related to the functionality of the program, such as screen resolution and color schemes. To accommodate these on-functional variations, researchers suggested fuzzy comparison rules that determine whether the output of a GUI program matches the oracles. Others suggested computer vision based solutions that make use of computationally expensive image processing techniques to abstract the strict comparisons. Alternatively, we propose GUICOP, a system that checks whether a trace of a GUI execution violates a given GUI specification. GUICOP is composed of a GUI specification language, instrumented GUI libraries, and a checker. The alphabet of the specification language contains basic geometric shapes describing GUI components, events, and positional and temporal operators that express relative object positions and event timings, respectively. During program execution, the instrumented libraries capture positional and temporal information of components and associated triggered events in execution traces. The checker determines whether the traces satisfy the specifications. To evaluate GUICOP, we wrote 50 use cases that describe real GUI applications and used the GUICOP checker on the supported cases that successfully revealed violations.