Format-Preserving Encryption (FPE) algorithms are symmetric cryptographic algorithms that encrypt an arbitrary-length plaintext into a ciphertext of the same size. Standardisation bodies recognised the first FPE algorithms (FEA-1, FEA-2, FF1 and FF3-1) in the last decade, and they have not been used for network layer privacy protection so far. However, their ability to encrypt arbitrary-length plaintext makes them suitable for encrypting selected packet header fields and replacing their original value with ciphertext of the same size without storing excessive information on the network element. If the encrypted fields carry personally identifiable information, it is possible to protect the privacy of the endpoints in the communication. This paper presents our research on using FPE for network layer privacy protection and describes LISPP, a lightweight, stateless network layer privacy protection system. The system was developed for programmable smart network interface cards (NIC) and thoroughly tested in a real network environment. We have created several implementations ranging from pure P4 to a mix of P4 and C implementations, exploring their performance and the suitability of target-independent P4 language for such processor-intensive applications. Finally, LISPP achieved line rate TCP throughput, up to 4.5 million packets per second, with the penalty of only 30 to 60 microseconds of additional one-way delay, proving that it is adequate for use in production networks. The most efficient implementation was with the FF3-1 algorithm developed in C and carefully adapted to the specific hardware configuration of the NIC.

Based on branch prediction, value prediction has emerged as a solution for problems caused by true data dependencies in pipelined processors. While branch predictors have binary outcomes (taken/not taken), value predictors face a challenging task as their outcomes can take any value. Because of that, coverage is reduced to enhance high accuracy and minimise costly recovery from misprediction. This paper evaluates value prediction, focusing on instruction execution with imprecisely predicted operands whose result can still be correct. Two analytical models are introduced to represent instruction execution with value prediction. One model focuses on correctly predicted operands, while the other allows for imprecisely predicted operands as long as the instruction results remain correct. A trace-driven simulator was developed for simulation purposes, implementing well-known predictors and some of the predictors presented at the latest Championship Value Prediction. The gem5 simulator was upgraded to generate program traces of SPEC and EEMBC benchmarks that were used in simulations. Based on the simulation result, proposed analytical models were compared to reveal the conditions under which a model with imprecisely predicted operands, but still correct results, achieved better execution time than a model with correctly predicted operands. Analysis revealed that the accuracy of the correct instruction result based on the predicted operand, even when the predicted operand is imprecise, is higher than the accuracy of the correctly predicted operand. The accuracy improvement ranges from 0.8% to 44%, depending on the specific predictor used.