At least 50 percent of all mobile device applications are free. These free apps and their associated ads may result in bandwidth consumption charges (needed to keep ads up to date) and battery drain (an application may poll for ad updates or embed ads in an application that accesses the cellular network) for the mobile user. Targeted ads in free apps can also result in a loss of privacy for the mobile user.
The authors suggest moving ad delivery to a cloudlet. A cloudlet is a micro instance running in a cloud environment for the purpose of posting new ads and acquiring information from the mobile device. It is assumed that the transmission overhead would be minimized because information related to a mobile device would be stored in the cloudlet.
An excellent argument is made for the efficiency of the cloudlet environment. Generating millions of cloudlets would create some pitfalls, however, with regard to security and privacy. If a cloudlet environment were compromised, this would be an excellent attack vector for transmitting malware into a mobile device. Furthermore, this paper was apparently written before the leaks concerning National Security Agency (NSA) surveillance of the Internet. Using cloudlets requires some kind of location information, which could provide enough data to create a profile of an anonymous mobile device user.
Because the authors assume that the cloudlets are anonymized, they do not discuss techniques that could be used to de-anonymize them and link them to their owners. Narayanan and Shmatikov’s paper provides a good discussion of de-anonymization [1].
The authors note that future work needs to be done with the cloudlet concept. They need to overcome some of the drawbacks associated with this idea before it can be introduced to the public in any kind of production environment.
The paper could have used some additional editing. There are numerous typographical and grammatical errors, which made it difficult to follow the discussion.