Profiling of symmetric-encryption algorithms for a novel biomedical-implant architecture

Abstract

Starting with the implantable pacemaker, microelectronic implants have been around for more than 50 years. A plethora of commercial and research-oriented devices have been developed so far for a wide range of biomedical applications. In view of an envisioned expanding implant market in the years to come, our ongoing research work is focusing on the specification and design of a novel biomedical microprocessor core, carefully tailored to a large subset of existing and future biomedical applications. Towards this end, we have taken steps in identifying various tasks commonly required by such applications and profiling their behavior and requirements. One such task is decryption of incoming commands to an implant and encryption of outgoing (telemetered) biological data. Secure bidirectional information relaying in implants has been largely overlooked so far although protection of personal (biological) data is very crucial. In this context, we evaluate a large number of symmetric (block) ciphers in terms of various metrics: average and peak power consumption, total energy budget, encryption rate and efficiency, program-code size and security level. For our study we use XTREM, a performance and power simulator for Intel's XScale embedded processor. Findings indicate the best-performing ciphers across most metrics to be MISTY1 (scores high in 5 out of 6 imposed metrics), IDEA and RC6 (both present in 4 out of 6 metrics). Further profiling of MISTY1 indicates a clear dominance of load/store, move and logic-operation instructions which gives us explicit directions for designing the architecture of our novel processor.

ImpBench revisited: An extended characterization of implant-processor benchmarks

“…5 further reveals that "misty" consumes less power but requires more area than "rc6" (and, thus, "all") while scoring better performance than either of them. This agrees with existing literature [23] and with the targeted applications of MISTY1 which are low-power, embedded systems. Its increased area requirements w.r.t.…”

“…EEG, EMG, blood pressure, pulmonary air volume). Without loss of generality, for this paper we have selected and executed only the 10 − KB EMG workload ("EMGII 10.bin", see [23] for details) as it exhibits worst-case performance characteristics and, thus, provides a lower-bound for processor design.…”

The case for a generic implant processor

“…So far, extensive work has been performed for identifying and profiling common applications to be executed on such an architecture. Algorithms for lossless data compression [24], symmetric-key encryption [25] and error detection/correction as well as representative real-world applications have been evaluated and suitable candidates have been isolated. Moreover, a carefully selected benchmark suite for microelectronic implants has been proposed [26] to guide and assist future implant design.…”

A system architecture, processor, and communication protocol for secure implants

“…Since modern IMDs are embedded computing systems in actuality, specially targeted computer viruses or malware could infect implantable devices and use them to spread, potentially damaging a large patient base [Gasson 2010]. Public awareness on the matter is gradually catching up [Leavitt 2010;Strydis et al 2008] and successful, in vitro attacks on implantable devices have already been demonstrated Li et al 2011] to verify that such fears are not unwarranted. The attacks on the Epilepsy Foundation website-where seizures were induced in some epileptic patients by replacing website content with rapidly flashing images [Poulson 2008]-have led to the certainty that malicious attacks on IMDs are simply a matter of time and, therefore, that security is required in order to protect the steadily increasing number of patients carrying wireless IMDs.…”