Toward Secure Data Computation and Outsource for Multi-User Cloud-Based IoT

Abstract : Cloud computing has promoted the success of Internet of Things (IoT) with offering abundant storage and computation resources where the data from IoT sensors can be remotely outsourced to the cloud servers, whereas storing, exchanging and processing data collected through IoT sensors via centralised or decentralised cloud servers make cloud-based IoT systems prone to internal or external attacks. To protect IoT data against potential malicious users and adversaries, some cryptographic schemes have been applied to ensure confidentiality and integrity of IoT data. It is however a challenging task to perform any arithmetical computations once data items are encrypted. Fully-homomorphic encryption which is based on lattices can, in principle, provide a solution, but it is unfortunately inefficient in computation and hence cannot be applied to IoT. Fully-homomorphic encryption is feasible when we allow an involvement of semi-trusted server. However, it is challenging to provide such a system in the situation of distributed environments for shared IoT data. We solve this problem and provide a fully-homomorphic encryption scheme for cloud-based IoT applications. We introduce a new method with the aid of semi-trusted server who can help in the computation of the homomorphic multiplications without gaining any useful information of the encrypted data.

 EXISTING SYSTEM :

 ? Majority of the existing SE schemes, including our previous work, are software-based solutions built on top of diverse cryptographic primitives, which result in a rich set of secure search indexes and algorithm designs. ? However, the existing secure deduplication designs, to some extent, are at odds with the real-world dedupe requirements in terms of security and performance. ? This gives us the desired asymmetry between security and performance, i.e. resilient to multiple compromised clients, compared to existing work, but only with minimal dedupe performance loss. ? Our scheme is on par with the plaintext practice in terms of the deduplication performance while gaining better security guarantees compared to the existing work.

 DISADVANTAGE :

 ? Cloud computing has fulfilled most of the demands of modern technology, it may not be a suitable solution as there are still unresolved problems, whereas IoT devices and applications need to be processed swiftly. ? As devices can always breakdown or become vulnerable to malicious attacks, authentication alone is not adequate to fix these problems. ? If some Fog nodes are compromised by any intruders, it is a problematic task to ensure the security of the data. ? Due to the resource constraints of the Fog devices, designing a high security and low cost threat and attack detection is the key problem in the Fog.

 PROPOSED SYSTEM :

 • The proposed SE scheme enables users to freely update the secure index and the corresponding file collection. • The proposed scheme incurs minimal ciphertext size expansion and computation overhead. • We demonstrate that the proposed system provides broad support for a variety of search functions and achieves computation efficiency comparable to plaintext data search with elevated security protection. • The security of the proposed scheme is derived from the MSSE security against adaptive chosen-keyword attacks. • The proposed system supports a rich set of IR functions and query types while ensuring the confidentiality and integrity of the query process.

 ADVANTAGE :

 ? Cloud computing also offers diverse features to users such as ease of access to information, cost efficiency, quick deployment, backup and recovery. ? However, CloudWatcher is unable to generate routing path and, if there many new flows in the network path, it is less efficient and performance degrades. ? Therefore, they need to use different authentication methods for different services where the performance of the authentication methods is different in the context of latency, efficiency and scalability. ? Security and performance are both highly required in terms of different contextual devices and applications. ? Bigdata are usually transferred over data transfer protocol and data transfer can be more efficient if we consider security since sensitive data are transferring over these protocols.