Ahmed Ahmed ahmed100553

📡 Research Identity

class RadarResearcher:
    """Signal model → experiment → real-time implementation → inference."""

    affiliation = "Advanced Radar Research Center | University of Oklahoma"
    degree      = "PhD Researcher, Electrical & Computer Engineering"

    research = {
        "radar": [
            "Detection, Estimation & Tracking",
            "Range-Doppler & Micro-Doppler Processing",
            "Coherent Processing & Phase Estimation",
            "Statistical / Higher-Order Signal Processing",
        ],
        "systems": [
            "Real-Time Radar Pipelines",
            "High-Rate Data Acquisition & Streaming",
            "CUDA / GPU-Accelerated DSP",
            "Hardware-Software Co-Design",
        ],
        "intelligence": [
            "Deep Learning for Radar",
            "Vision Transformers",
            "Target Characterization & Classification",
        ],
    }

    current_mission = (
        "Build experimental real-time radar systems for aerial-target sensing, "
        "then develop algorithms that survive real hardware, real clutter, and real latency."
    )

Research philosophy: an algorithm is most interesting when it survives the complete chain — measurement physics, acquisition, synchronization, real-time computation, detection, and experimental validation.

⚡ Current Mission — Real-Time Aerial Target Radar

┌──────────────┐    ┌────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│ RADAR SENSOR │───▶│ IF / IQ STREAM │───▶│ BUFFER + SYNC   │───▶│ GPU DSP         │
│ measurement  │    │ high-rate data │    │ bounded latency │    │ FFT / filtering │
└──────────────┘    └────────────────┘    └─────────────────┘    └────────┬────────┘
                                                                          │
                                                                          ▼
┌────────────────┐    ┌────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│ TARGET INTEL.  │◀───│ μ-DOPPLER / ML │◀───│ TRACKING        │◀───│ CFAR / DETECTION│
│ decision layer │    │ characterization│    │ state estimation│    │ range–Doppler   │
└────────────────┘    └────────────────┘    └─────────────────┘    └─────────────────┘

Goal: bridge advanced radar algorithms with reliable, low-latency experimental systems — not offline demos alone.

🎯 Research Contributions

📡 Coherent-On-Receive Processing

Software-defined phase recovery for magnetron-based marine radar

Metric	Achievement
Phase RMSE	< 0.01 rad
Channel Isolation	94–113 dB
Collaboration	PNNL & USGS

Key idea: extract a stable transmit-leakage phase reference so a nominally non-coherent radar can support coherent Doppler processing.

🤖 FAN / Vision Transformer Radar AI

Compact CNN–Transformer architecture for micro-Doppler classification

Metric	Achievement
Accuracy	98%
Parameters	1.8M
Parameter reduction	98%

Key idea: Fourier-aware neural layers for efficient learning of periodic structure in radar time-frequency signatures.

📊 HSSD Framework

Hybrid Statistic Subspace Decomposition

4th-Order Cumulants
        +
       SVD
        +
    Shrinkage
        │
        ▼
Non-Gaussian Component Isolation
        │
        ▼
Clutter / Leakage Suppression

Key idea: replace conventional covariance structure with a fourth-order cumulant representation to emphasize non-Gaussian components.

🫀 Non-Contact Vital Signs

FMCW/mmWave radar for heart- and breathing-rate estimation

Validation	Scale
Participants	50 children
Measurements	250 min
Radar frames	25,000

Key idea: dual signal-processing back ends — fixed Butterworth filtering and adaptive MODWT separation — for robust physiological extraction.

🔬 Radar Mathematics I Work With

┌────────────────────────────────────────────────────────────────────────────────────────────┐
│                                  RADAR PROCESSING CORE                                     │
├────────────────────────────────────────────────────────────────────────────────────────────┤
│                                                                                            │
│  Measurement model                                                                        │
│  x[n,m] = Σₚ αₚ s[n-τₚ] exp(j2π fD,ₚ mTr + jφₚ) + w[n,m]                                │
│                                                                                            │
│  Range–Doppler transform                                                                  │
│  XRD[k,ℓ] = Σₘ Σₙ x[n,m] wr[n] wd[m] exp(-j2πkn/N) exp(-j2πℓm/M)                         │
│                                                                                            │
│  CFAR threshold                                                                           │
│  T[k,ℓ] = αCFAR · (1/Ntrain) Σ(i,j)∈𝒯 |XRD[k+i,ℓ+j]|²                                    │
│                                                                                            │
│  Micro-Doppler / local slow-time spectrum                                                 │
│  S(t,f) = Σq z[q] g[q-t] exp(-j2πfq)                                                      │
│                                                                                            │
└────────────────────────────────────────────────────────────────────────────────────────────┘

measurement → estimation → transform → detection → tracking → inference

🛠️ Technical Arsenal

Radar • DSP • Numerical Computing

Real-Time • GPU • Systems

Machine Learning • Deep Learning

Radar Hardware • Edge • Infrastructure

🚀 Selected Public Projects

🫀 VitalFlow-Radar-Public

77 GHz mmWave radar + edge DSP + real-time streaming + cloud AI for contactless vital-sign monitoring.

Radar DSP Kafka Vertex AI Real-Time Dashboard

📈 FMCW-HR-BR-Estimation

End-to-end 60 GHz FMCW radar pipeline for non-contact heart- and breathing-rate estimation.

MATLAB FMCW MODWT Biomedical Radar

🩻 OCT-Stripes-Removal

Frequency-domain image processing for suppression of stripe artifacts in optical coherence tomography imagery.

FFT OpenCV Image Processing

🤖 MazeSolvingDQN

Deep Q-learning experiments for sequential decision-making and autonomous maze navigation.

Reinforcement Learning DQN Python

Some of my current radar systems and experimental research repositories are private or collaboration-restricted.

🏛️ Research Experience

Position	Institution	Focus	Duration
Graduate Research Assistant	Advanced Radar Research Center	Radar DSP, experimental systems, ML	May 2024 – Present
Undergraduate Research Assistant	Advanced Radar Research Center	Sensor fusion, ROS, interference mitigation	Jul 2022 – May 2024
Undergraduate Researcher	3SigmaLab	BCI, EEG signal processing, ML	Jan 2022 – May 2022

🧭 Research Trajectory

      SIGNAL PROCESSING                    REAL-TIME SYSTEMS                   RADAR INTELLIGENCE
             │                                    │                                  │
             ▼                                    ▼                                  ▼
  statistical estimation  ───────▶  high-rate acquisition / GPU  ───────▶  detection / tracking / ML
             │                                    │                                  │
             └────────────────────────── experimental validation ────────────────────┘

I am interested in radar research where theory, algorithms, hardware, and implementation meet — especially problems involving aerial targets, cluttered environments, low-latency processing, and deployable sensing systems.

📬 Connect

"The theory is the map. The measurement is the territory."

Navigation Menu