CEVA-LG全栈UWB平台：TSMC 22nm硅验证与802.15.4ab CPD落地路径

import numpy as np
from typing import Tuple

def simulate_mms_ranging(
    n_pulse_segments: int = 10,
    segment_duration_us: float = 100,
    interval_us: float = 1000,
    psd_limit_dbm_mhz: float = -41.3,
    bandwidth_mhz: float = 500
) -> Tuple[float, float, float]:
    """
    模拟MMS多毫秒测距的累计能量增益
    
    Args:
        n_pulse_segments: MMS分段数
        segment_duration_us: 每段持续时间(微秒)
        interval_us: 段间间隔(微秒)
        psd_limit_dbm_mhz: PSD限制(dBm/MHz)
        bandwidth_mhz: 信号带宽(MHz)
    
    Returns:
        (single_energy_dbm, mms_energy_dbm, gain_db)
    
    参考: IEEE 802.15.4ab MMS机制
    """
    # 单脉冲能量 (dBm)
    single_power_dbm = psd_limit_dbm_mhz + 10 * np.log10(bandwidth_mhz)
    
    # MMS累计能量: N段相干累加
    # 相干累加增益 = 10*log10(N)
    mms_energy_dbm = single_power_dbm + 10 * np.log10(n_pulse_segments)
    
    gain_db = mms_energy_dbm - single_power_dbm
    
    return single_power_dbm, mms_energy_dbm, gain_db


def compare_ranging_standards() -> dict:
    """
    对比802.15.4z与802.15.4ab测距能力
    
    Returns:
        对比字典
    """
    results = {}
    
    # 802.15.4z: 单脉冲
    single, _, _ = simulate_mms_ranging(n_pulse_segments=1)
    results['802.15.4z_single_pulse'] = {
        'energy_dbm': single,
        'max_range_m': 30,  # 典型值
        'nlos_capable': False
    }
    
    # 802.15.4ab MMS: 10段
    _, mms10, gain10 = simulate_mms_ranging(n_pulse_segments=10)
    results['802.15.4ab_mms_10seg'] = {
        'energy_dbm': mms10,
        'gain_db': gain10,
        'estimated_max_range_m': 30 * np.sqrt(10),  # 能量平方根关系
        'nlos_capable': True
    }
    
    # 802.15.4ab MMS: 30段 (CEVA声称30x)
    _, mms30, gain30 = simulate_mms_ranging(n_pulse_segments=30)
    results['802.15.4ab_mms_30seg'] = {
        'energy_dbm': mms30,
        'gain_db': gain30,
        'estimated_max_range_m': 30 * 30,  # CEVA声称值
        'nlos_capable': True
    }
    
    return results


if __name__ == "__main__":
    print("=== 802.15.4z vs 802.15.4ab 测距能力对比 ===\n")
    
    results = compare_ranging_standards()
    for std, data in results.items():
        print(f"[{std}]")
        for k, v in data.items():
            print(f"  {k}: {v:.2f}" if isinstance(v, float) else f"  {k}: {v}")
        print()
    
    # MMS增益计算
    single, mms, gain = simulate_mms_ranging(n_pulse_segments=10)
    print(f"MMS 10段累加增益: {gain:.1f} dB")
    print(f"单脉冲能量: {single:.2f} dBm")
    print(f"MMS累计能量: {mms:.2f} dBm")
    
    # 30段
    _, mms30, gain30 = simulate_mms_ranging(n_pulse_segments=30)
    print(f"\nMMS 30段累加增益: {gain30:.1f} dB")
    print(f"MMS 30段累计能量: {mms30:.2f} dBm")

# CEVA-LG UWB集成到IMS的参考架构
class UWBCPDSystem:
    """
    UWB CPD系统集成参考架构
    
    基于CEVA-Waves 802.15.4ab全栈IP
    """
    
    def __init__(self, config: dict):
        self.standard = config.get('standard', '802.15.4ab')
        self.channel = config.get('channel', 11)  # 1.3GHz带宽
        self.mms_segments = config.get('mms_segments', 30)
        self.ranging_mode = config.get('ranging_mode', 'MMS+NBA')
        self.cpd_threshold = config.get('cpd_threshold', 0.7)
        
    def detect_presence(self, csi_data: np.ndarray) -> dict:
        """
        基于CSI数据进行CPD检测
        
        Args:
            csi_data: 信道状态信息, shape=(n_subcarriers, n_samples)
        
        Returns:
            检测结果字典
        """
        # Step 1: PCA降维
        from sklearn.decomposition import PCA
        pca = PCA(n_components=5)
        components = pca.fit_transform(csi_data.T)
        
        # Step 2: 自适应主成分选择
        # 基于Welch功率谱密度
        from scipy.signal import welch
        scores = []
        for i in range(components.shape[1]):
            f, psd = welch(components[:, i], fs=100, nperseg=256)
            # 人体运动频段: 0.5-3 Hz (呼吸/心跳)
            motion_band = np.sum(psd[(f >= 0.5) & (f <= 3.0)])
            total_power = np.sum(psd)
            score = motion_band / total_power
            scores.append(score)
        
        best_component = np.argmax(scores)
        
        # Step 3: 小波变换检测微动
        import pywt
        coeffs = pywt.cwt(components[:, best_component], 
                          scales=np.arange(1, 50), 
                          wavelet='morl')
        
        # Step 4: 能量阈值判定
        energy = np.abs(coeffs[0])
        detection_score = np.mean(energy > self.cpd_threshold)
        
        return {
            'presence_detected': detection_score > 0.5,
            'confidence': float(detection_score),
            'best_component': int(best_component),
            'motion_band_score': float(scores[best_component]),
            'standard': self.standard,
            'channel': self.channel
        }