Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Akhilesh Tyagi has a total of 27 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Large Pages, Large Leaks? Hugepage-Induced Side-Channels vs. Performance Improvements in Cryptographic Computations

    by Xinyao Li, Akhilesh Tyagi

    Side-channel attacks leveraging microarchitectural components such as caches and translation lookaside buffers (TLBs) pose increasing risks to cryptographic and machine-learning workloads. This paper presents a comparative study of performance and side-channel leakage under two page-size configurations—standard 4 KB pages and 2 MB huge pages—using paired attacker–victim experiments instrumented with both Performance Monitoring Unit (PMU) counters and precise per-access timing using rdtscp(). The victim executes repeated, key-dependent memory accesses across eight cryptographic modes (AES, ChaCha20, RSA, and ECC variants) while the attacker records eight PMU features per access (cpu-cycles, instructions, cache-references, cache-misses, etc.) and precise rdtscp() timing. The resulting traces are analyzed using a multilayer perceptron classifier to quantify key-dependent leakage. Results show that the 2 MB huge-page configuration achieves a comparable key-classification accuracy (mean 0.79 vs. 0.77 for 4 KB) while reducing average CPU cycles by approximately 11%. Page-index identification remains near random chance (3.6–3.7% for PMU side-channels and 1.5% for timing side-channel), indicating no increase in measurable leakage at the page level. These findings suggest that huge-page mappings can improve runtime efficiency without amplifying observable side-channel vulnerabilities, offering a practical configuration for balancing performance and security in user-space cryptographic workloads.


  • Open Access

    Decoupled, Real-Time, Multi-Modal Program Execution State Monitoring

    by Ananda Biswas, Dakota Berbrich, Braedon Giblin, Zelong Li, Joyesh Philip, Ravikumar Selvam, Henry Duwe, Akhilesh Tyagi

    Many layer PCBs have many attack surfaces that can hide a Trojan capable of corrupting the processor execution state only with interactions with the processor through external channels such as memory bus. The focus of this research is to monitor the processor execution state only through channels and side-channels that extend beyond the processor chip boundaries. Such a decoupled monitor localizes the program execution state at a blob level - an aggregated form of the program control flow graph. Higher the level of blob aggregation, less demanding are the requirements for the side-channels and execution state revealing channels. Decoupled monitor uses side-channel sensor streams that are naturally created by the program execution (last level cache -LLC- misses). The side-channel sensor streams evaluated in this paper are (1) LLC miss address stream, (2) processor domain power stream, (3) DDR memory domain power stream captured through electromagnetic (EM) emission, and (4) performance monitoring unit (PMU) stream. Blob construction heuristics presents a monitoring overhead trade-off with the localization granularity. A blob is a program level entity whose boundaries are detectable off-processor through side-channel streams. Typical blob sizes we have encountered are 200 instructions as static size and 100s of millions of dynamically executed instructions. The goal of the decoupled monitor is to validate the execution state conformity with a precomputed golden model at a blob and blob path granularity. The monitor is evaluated on a Xilinx Zynq Ultrascale+ ZCU 106 board which contains two ARM processors and a sea of FPGA fabric. Targeted program executes on the Cortex A53 processor for the monitored program state localization. Each of the LLC address, execution path power, performance monitoring unit streams builds machine learning (ML) models for all the paths in a program. The monitor uses these trained ML models to classify the sensor stream data into a Blob/Path. The multiple streams’ classifications are resolved into a single Blob/Path localization based on confidence values of each stream classification. Individual stream’s classification accuracy ranges from 80-90% for the Blob/Path classification. The overall execution state localization is evaluated on a benchmark program "STREAM" with 3 normal execution runs and 2 anomalous runs. The accuracy of this localization is 83.3% for normal runs and 100% for anomalous runs.

  • Open Access

    Large Pages, Large Leaks? Hugepage-Induced Side-Channels vs. Performance Improvements in Cryptographic Computations

    by Xinyao Li, Akhilesh Tyagi

    Side-channel attacks leveraging microarchitectural components such as caches and translation lookaside buffers (TLBs) pose increasing risks to cryptographic and machine-learning workloads. This paper presents a comparative study of performance and side-channel leakage under two page-size configurations—standard 4KB pages and 2MB huge pages—using paired attacker–victim experiments instrumented with both Performance Monitoring Unit (PMU) counters and precise per-access timing using rdtscp(). The victim executes repeated, key-dependent memory accesses across eight cryptographic modes (AES, ChaCha20, RSA, and ECC variants) while the attacker records eight PMU features per access (cpu-cycles, instructions, cache-references, cache-misses, etc.) and precise rdtscp() timing. The resulting traces are analyzed using a multilayer perceptron classifier to quantify key-dependent leakage. Results show that the 2MB huge-page configuration achieves a comparable key-classification accuracy (mean 0.79 vs. 0.77 for 4KB) while reducing average CPU cycles by approximately 11%. Page-index identification remains near random chance (3.6--3.7% for PMU side-channels and 1.5% for timing side-channel), indicating no increase in measurable leakage at the page level. These findings suggest that huge-page mappings can improve runtime efficiency without amplifying observable side-channel vulnerabilities, offering a practical configuration for balancing performance and security in user-space cryptographic workloads.

  • Open Access

    Multi-Line Prefetch Covert Channel with Huge Pages

    by Xinyao Li, Akhilesh Tyagi

    Modern x86 processors incorporate performance-enhancing features such as prefetching mechanisms, cache coherence protocols, and support for large memory pages (e.g., 2 MB huge pages). While these architectural innovations aim to reduce memory access latency, boost throughput, and maintain cache consistency across cores, they can also expose subtle microarchitectural side channels that adversaries may exploit. This study investigates how the combination of prefetching techniques and huge pages can significantly enhance the throughput and accuracy of covert channels in controlled computing environments. Building on prior work that examined the impact of the MESI cache coherence protocol using single-cache-line access without huge pages, our approach expands the attack surface by simultaneously accessing multiple cache lines across all 512 L1 lines under a 2 MB huge page configuration. As a result, our 9-bit covert channel achieves a peak throughput of 4940 KB/s—substantially exceeding previously reported benchmarks. We further validate our channel on AMD SEV-SNP virtual machines, achieving up to an 88% decoding accuracy using write-access encoding with 2 MB huge pages, demonstrating feasibility even under TEE-enforced virtualization environments. These findings highlight the need for careful consideration and evaluation of the security implications of common performance optimizations with respect to their side-channel potential.


  • Open Access

    Microarchitecture Evaluation Framework for Transient Execution Attack Vulnerability: Metrics, Fuzzing, and Sensitivity Analysis

    by Jordan McGhee, Nayra Lujano, Aiden Peterson, Henry Duwe, Akhilesh Tyagi, Berk Gulmezoglu

  • Open Access

    Residue Number System (RNS) and Power Distribution Network Topology-Based Mitigation of Power Side-Channel Attacks

    by Ravikumar Selvam, Akhilesh Tyagi

    Over the past decade, significant research has been performed on power side-channel mitigation techniques. Logic families based on secret sharing schemes, such as t-private logic, that serve to secure cryptographic implementations against power side-channel attacks represent one such countermeasure. These mitigation techniques are applicable at various design abstraction levels—algorithm, architecture, logic, physical, and gate levels. One research question is when can the two mitigation techniques from different design abstraction levels be employed together gainfully? We explore this notion of the orthogonality of two mitigation techniques with respect to the RNS secure logic, a logic level power side-channel mitigation technique, and power distribution network (PDN), with the decoupling capacitance, a mitigation technique at physical level. Machine learning (ML) algorithms are employed to measure the effectiveness of power side-channel attacks in terms of the success rate of the adversary. The RNS protected LED block cipher round function is implemented as the test circuit in both tree-style and grid-style PDN using the FreePDK 45 nm technology library. The results show that the success rate of an unsecured base design 68.96% for naive Bayes, 67.44% with linear discriminant analysis, 67.51% for quadratic discriminant analysis, and 66.58% for support vector machine. It is reduced to a success rate of 19.68% for naive Bayes, 19.62% with linear discriminant analysis, 19.10% for quadratic discriminant analysis, and 10.54% in support vector machine. Grid-type PDN shows a slightly better reduction in success rate compared to the tree-style PDN.


  • Open Access

    Block-Active ADMM to Minimize NMF with Bregman Divergences

    by Xinyao Li, Akhilesh Tyagi

    Over the last ten years, there has been a significant interest in employing nonnegative matrix factorization (NMF) to reduce dimensionality to enable a more efficient clustering analysis in machine learning. This technique has been applied in various image processing applications within the fields of computer vision and sensor-based systems. Many algorithms exist to solve the NMF problem. Among these algorithms, the alternating direction method of multipliers (ADMM) and its variants are one of the most popular methods used in practice. In this paper, we propose a block-active ADMM method to minimize the NMF problem with general Bregman divergences. The subproblems in the ADMM are solved iteratively by a block-coordinate-descent-type (BCD-type) method. In particular, each block is chosen directly based on the stationary condition. As a result, we are able to use much fewer auxiliary variables and the proposed algorithm converges faster than the previously proposed algorithms. From the theoretical point of view, the proposed algorithm is proved to converge to a stationary point sublinearly. We also conduct a series of numerical experiments to demonstrate the superiority of the proposed algorithm.


  • Open Access

    Cross-World Covert Channel on ARM Trustzone through PMU

    by Xinyao Li, Akhilesh Tyagi

    The TrustZone technology is incorporated in a majority of recent ARM Cortex A and Cortex M processors widely deployed in the IoT world. Security critical code execution inside a so-called secure world is isolated from the rest of the application execution within a normal world. It provides hardware-isolated area called a trusted execution environment (TEE) in the processor for sensitive data and code. This paper demonstrates a vulnerability in the secure world in the form of a cross-world, secure world to normal world, covert channel. Performance counters or Performance Monitoring Unit (PMU) events are used to convey the information from the secure world to the normal world. An encoding program generates appropriate PMU event footprint given a secret S. A corresponding decoding program reads the PMU footprint and infers S using machine learning (ML). The machine learning model can be trained entirely from the data collected from the PMU in user space. Lack of synchronization between PMU start and PMU read adds noise to the encoding/decoding ML models. In order to account for this noise, this study proposes three different synchronization capabilities between the client and trusted applications in the covert channel. These are synchronous, semi-synchronous, and asynchronous. Previously proposed PMU based covert channels deploy L1 and LLC cache PMU events. The latency of these events tends to be 100–1000 cycles limiting the bandwidth of these covert channels. We propose to use microarchitecture level events with latency of 10–100 cycles captured through PMU for covert channel encoding leading to a potential 100× higher bandwidth. This study conducts a series of experiments to evaluate the proposed covert channels under various synchronization models on a TrustZone supported Cortex-A processor using OP-TEE framework. As stated earlier, switch from signaling based on PMU cache events to PMU microarchitectural events leads to approximately 15× higher covert channel bandwidth. This proposed finer-grained microarchitecture event encoding covert channel can achieve throughput of the order of 11 Kbits/s as opposed to previous work’s throughput of the order of 760 bits/s.


  • Open Access

    Continuous Nonintrusive Mobile Device Soft Keyboard Biometric Authentication

    by Timothy Dee, Ian Richardson, Akhilesh Tyagi

    Mobile banking, shopping, and in-app purchases utilize persistent authentication states for access to sensitive data. One-shot authentication permits access for a fixed time period. For instance, a username/password-based authentication allows a user access to all the shopping and payments data in the Amazon shopping app. Traditional user passwords and lock screens are easily compromised. Snooping attacks—observing an unsuspecting user entering passwords—and smudge attacks—examining touchscreen finger oil residue—enable compromised user authentication. Mobile device interactions provide robust human and device identity data. Such biometrics enhance authentication. In this paper, behavioral attributes during user input constitute the password. Adversary password reproduction difficulty increases since pure observation is insufficient. Current mobile continuous authentication schemes use, among others, touchscreen–swipe interactions or keyboard input timing. Many of these methods require cumbersome training or intrusive authentication. Software keyboard interactions provide a consistent biometric data stream. We develop biometric profiles using touch pressure, location, and timing. New interactions authenticate against a profile using a divergence measure. In our limited user–device data sets, the classification achieves virtually perfect accuracy.