With the development of Internet of Things (IoT) technology, smart mobile devices are widely used in daily life. The service providers always extensively collect data from users for training machine learning models in order to improve their accurate services. This also raises users’ concerns about data privacy and security. Federated learning, as an extension of centralized machine learning, allows several users working together to train a machine learning model on their own devices without sending their data to the centralized servers. However, existing research suggests that local models also contain privacy related to the users’ data. Unfortunately, the current privacy-preserving secure aggregation methods have either poor accuracy or high computational and communication costs in training process which can not afford by the IoT devices. In this work, we propose a federated learning framework supporting privacy-preserving data aggregation against external and internal attackers with lower computational and communication costs, which is suitable for the weak IoT devices. The scheme is also supporting aggregation with fault tolerance and dynamic user set even if a part of users leave the system in the training. Detailed security analysis and extensive experiments using a real dataset confirm the efficacy and efficiency of the proposed schemes.

Homomorphic encryption always allows the linear arithmetic operations performed over the ciphertext and then returns equaling results as if the operations are taken over the original plaintext, which is always used for data aggregation in wireless sensor networks to keep the confidentiality of the data and cut down the transmission overhead of the ciphertext. In the marine sensor networks, sensors collect the multiple data such as temperature, salinity, pressure, and chlorophyll concentration in the ocean using a single hardware unit for further statistical analysis such as computing the mean and the variance and making regression analysis. However, directly using the homomorphic encryption cannot perform well in marine sensor data forwarding since the data need to turn to satellites or vessels as relays and be forwarded in multi-hop way. The data are not expected to be decrypted until arriving the final destinations. To tackle these issues, we design a secure data forwarding protocol based on the Paillier homomorphic encryption and multi-use proxy re-encryption. We also evaluate the computational overhead in term of the delay in the transmission and operation in various test beds. The experiment results show that the additional computational overhead brought by cryptographic operations could be minor and it has the merit of providing fixed data size passing through the multi-hop transmission.

Since the technology of mobile cloud computing has brought a lot of benefits to information world, many applications in mobile devices based on cloud have emerged and boomed in the last years. According to the storage limitation, data owners would like to upload and further share the data through the cloud. Due to the safety requirements, mobile data owners are requested to provide credentials such as authentication tags along with the data. However, it is impossible to require mobile data owners to provide every authenticated computational results. The solution that signers’ privilege is outsourced to the cloud would be a promising way. To solve this problem, we propose three secure multi-entities delegated authentication protocols (MEDAPs) in mobile cloud computing, which enables the multiple mobile data owners to authorize a group designated cloud servers with the signing rights. The security of MEDAPs is constructed on three cryptographic primitive identity-based multi-proxy signature (IBMPS), identity-based proxy multi-signature (IBPMS), and identity-based multi-proxy multi-signature (IBMPMS), relied on the cubic residues, equaling to the integer factorization assumption. We also give the formal security proof under adaptively chosen message attacks and chosen identity/warrant attacks. Furthermore,compared with the pairing based protocol, MEDAPs are quite efficient and the communication overhead is nearly not a linear growth with the number of cloud servers. Copyright⃝c 2015 John Wiley & Sons, Ltd.