This conference proceedings publication is the result of a merge of two independent and concurrent works. The two papers were authored by Goldwasser, Goyal, Jain, and Sahai; and by Gordon, Katz, Liu, Shi, and Zhou. Research supported by NSFEAGER award # CNS1347364 DARPA award # FA8750-11-2-0225 and the Simons Foundation -Investigation Award. Research supported by NSF awards #1111599 and #1223623, and by the US Army Research Laboratory and the UK Ministry of Defence under Agreement Number W911NF-06-3-0001. The views and conclusions contained herein are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the US Army Research Laboratory, the U.S. Government, the UK Ministry of Defense, or the UK Government. The US and UK Governments are authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation hereon. Multi-input Functional Encryption 579Abstract. We introduce the problem of Multi-Input Functional Encryption, where a secret key sk f can correspond to an n-ary function f that takes multiple ciphertexts as input. We formulate both indistinguishability-based and simulation-based definitions of security for this notion, and show close connections with indistinguishability and virtual black-box definitions of obfuscation. Assuming indistinguishability obfuscation for circuits, we present constructions achieving indistinguishability security for a large class of settings. We show how to modify this construction to achieve simulationbased security as well, in those settings where simulation security is possible.

It is notoriously difficult to create hardware that is immune from side channel and tampering attacks. A lot of recent literature, therefore, has instead considered algorithmic defenses from such attacks.In this paper, we show how to algorithmically secure any cryptographic functionality from continual split-state leakage and tampering attacks. A split-state attack on cryptographic hardware is one that targets separate parts of the hardware separately. Our construction does not require the hardware to have access to randomness. In contrast, prior work on protecting from continual combined leakage and tampering [28] required true randomness for each update. Our construction is in the common reference string (CRS) model; the CRS must be hard-wired into the device. We note that prior negative results show that it is impossible to algorithmically secure a cryptographic functionality against a combination of arbitrary continual leakage and tampering attacks without true randomness; therefore restricting our attention to the split-state model is justified.Our construction is simple and modular, and relies on a new construction, in the CRS model, of non-malleable codes with respect to split-state tampering functions, which may be of independent interest.

Abstract. We consider the problem of delegating computation, where the delegator doesn't even know the input to the function being delegated, and runs in time significantly smaller than the input length.For example, consider the setting of memory delegation, where a delegator wishes to delegate her entire memory to the cloud. The delegator may want the cloud to compute functions on this memory, and prove that the functions were computed correctly. As another example, consider the setting of streaming delegation, where a stream of data goes by, and a delegator, who cannot store this data, delegates this task to the cloud. Later the delegator may ask the cloud to compute statistics on this streaming data, and prove the correctness of the computation. We note that in both settings the delegator must keep a (short) certificate of the data being delegated, in order to later verify the correctness of the computations. Moreover, in the streaming setting, this certificate should be computed in a streaming manner.We construct both memory and streaming delegation schemes. We present non-interactive constructions based on the (standard) delegation scheme of Goldwasswer et. al. [GKR08]. These schemes allow the delegation of any function computable by an L-uniform circuit of low depth (the complexity of the delegator depends linearly on the depth). For memory delegation, we rely on the existence of a polylog PIR scheme, and for streaming, we rely on the existence of a fully homomorphic encryption scheme.We also present constructions based on the CS-proofs of Micali. These schemes allow the delegation of any function in P. However, they are interactive (i.e., consists of 4 messages), or are non-interactive in the Random Oracle Model.

At TCC 2013, Choi et al. introduced the notion of multiclient verifiable computation (MVC) in which a set of clients outsource to an untrusted server the computation of a function f over their collective inputs in a sequence of time periods. In that work, the authors defined and realized multi-client verifiable computation satisfying soundness against a malicious server and privacy against the semi-honest corruption of a single client. Very recently, Goldwasser et al. (Eurocrypt 2014) provided an alternative solution relying on multi-input functional encryption. Here we conduct a systematic study of MVC, with the goal of satisfying stronger security requirements. We begin by introducing a simulationbased notion of security that provides a unified way of defining soundness and privacy, and automatically captures several attacks not addressed in previous work. We then explore the feasibility of achieving this notion of security. Assuming no collusion between the server and the clients, we demonstrate a protocol for multi-client verifiable computation that achieves stronger security than the protocol of Choi et al. in several respects. When server-client collusion is possible, we show (somewhat surprisingly) that simulation-based security cannot be achieved, even assuming only semi-honest behavior.