实验结果与结论

scripts/plot_thpt.py

@@ -0,0 +1,163 @@
+from __future__ import annotations
+
+import argparse
+import csv
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import numpy as np
+from matplotlib import font_manager
+
+
+BASE_FONT_SIZE = 25
+TITLE_FONT_SIZE = BASE_FONT_SIZE + 1
+AXIS_LABEL_FONT_SIZE = BASE_FONT_SIZE
+TICK_FONT_SIZE = BASE_FONT_SIZE - 2
+LEGEND_FONT_SIZE = BASE_FONT_SIZE - 2
+BAR_LABEL_FONT_SIZE = BASE_FONT_SIZE - 4
+MAX_LOSS = 0.02
+
+
+def configure_fonts() -> None:
+    preferred_fonts = [
+        "Microsoft YaHei",
+        "SimHei",
+        "Noto Sans CJK SC",
+        "Source Han Sans SC",
+        "PingFang SC",
+        "WenQuanYi Micro Hei",
+        "Arial Unicode MS",
+    ]
+    installed_fonts = {font.name for font in font_manager.fontManager.ttflist}
+    for font in preferred_fonts:
+        if font in installed_fonts:
+            plt.rcParams["font.sans-serif"] = [font]
+            break
+    plt.rcParams["axes.unicode_minus"] = False
+    plt.rcParams.update(
+        {
+            "font.size": BASE_FONT_SIZE,
+            "axes.titlesize": TITLE_FONT_SIZE,
+            "axes.labelsize": AXIS_LABEL_FONT_SIZE,
+            "xtick.labelsize": TICK_FONT_SIZE,
+            "ytick.labelsize": TICK_FONT_SIZE,
+            "legend.fontsize": LEGEND_FONT_SIZE,
+            "figure.dpi": 150,
+        }
+    )
+
+
+def read_throughput(csv_path: Path) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+    losses: list[float] = []
+    cubic: list[float] = []
+    fec: list[float] = []
+
+    with csv_path.open("r", encoding="utf-8-sig", newline="") as file:
+        reader = csv.DictReader(file)
+        for row in reader:
+            losses.append(float(row["loss"]))
+            cubic.append(float(row["cubic_thpt_mbps"]))
+            fec.append(float(row["fec_thpt_mbps"]))
+
+    return np.array(losses), np.array(cubic), np.array(fec)
+
+
+def format_loss_label(loss: float) -> str:
+    if loss == 0:
+        return "0"
+    percent = loss * 100
+    return f"{percent:g}%"
+
+
+def filter_results(
+    losses: np.ndarray,
+    cubic: np.ndarray,
+    fec: np.ndarray,
+) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+    keep = losses <= MAX_LOSS
+    return losses[keep], cubic[keep], fec[keep]
+
+
+def plot_absolute_throughput(losses: np.ndarray, cubic: np.ndarray, fec: np.ndarray, output_path: Path) -> None:
+    labels = [format_loss_label(loss) for loss in losses]
+
+    fig, ax = plt.subplots(figsize=(5.6, 5.0), constrained_layout=True)
+
+    ax.plot(losses, cubic, marker="o", linewidth=2.4, markersize=7, label="直接转发", color='#4285F4')
+    ax.plot(losses, fec, marker="s", linewidth=2.4, markersize=7, label="本文方法", color='#EA4335')
+    ax.set_xlabel("丢包率")
+    ax.set_ylabel("吞吐量（Mbps）")
+    ax.set_xticks(losses, labels)
+    plt.setp(ax.get_xticklabels(), rotation=-35, ha="left", rotation_mode="anchor")
+    ax.grid(axis="y", linestyle="--", alpha=0.35)
+    ax.grid(axis="x", linestyle="--", alpha=0.22)
+    ax.legend(frameon=False)
+
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(output_path, bbox_inches="tight")
+    plt.close(fig)
+
+
+def plot_speedup(losses: np.ndarray, cubic: np.ndarray, fec: np.ndarray, output_path: Path) -> None:
+    speedup = fec / cubic
+    labels = [format_loss_label(loss) for loss in losses]
+    category_x = np.arange(len(labels))
+
+    fig, ax = plt.subplots(figsize=(5.6, 5.0), constrained_layout=True)
+
+    bars = ax.bar(category_x, speedup, color="#4285F4", width=0.62)
+    ax.axhline(1.0, color="#444444", linewidth=1.2, linestyle="--")
+    ax.set_xlabel("丢包率")
+    ax.set_ylabel("相对吞吐提升")
+    ax.set_xticks(category_x, labels)
+    ax.grid(axis="y", linestyle="--", alpha=0.35)
+    ax.set_ymargin(0.1)
+
+    for bar, value in zip(bars, speedup):
+        ax.text(
+            bar.get_x() + bar.get_width() / 2,
+            bar.get_height(),
+            f"{value:.1f}x",
+            ha="center",
+            va="bottom",
+            fontsize=BAR_LABEL_FONT_SIZE,
+        )
+
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(output_path, bbox_inches="tight")
+    plt.close(fig)
+
+
+def plot_throughput(csv_path: Path, absolute_output_path: Path, speedup_output_path: Path) -> None:
+    configure_fonts()
+    losses, cubic, fec = read_throughput(csv_path)
+    losses, cubic, fec = filter_results(losses, cubic, fec)
+    plot_absolute_throughput(losses, cubic, fec, absolute_output_path)
+    plot_speedup(losses, cubic, fec, speedup_output_path)
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Plot throughput comparison under different loss rates.")
+    parser.add_argument("--input", type=Path, default=Path("scripts/thpt.csv"), help="Path to throughput CSV.")
+    parser.add_argument(
+        "--absolute-output",
+        type=Path,
+        default=Path("figures/thpt_absolute.pdf"),
+        help="Output path for the absolute throughput figure.",
+    )
+    parser.add_argument(
+        "--speedup-output",
+        type=Path,
+        default=Path("figures/thpt_speedup.pdf"),
+        help="Output path for the relative speedup figure.",
+    )
+    return parser.parse_args()
+
+
+def main() -> None:
+    args = parse_args()
+    plot_throughput(args.input, args.absolute_output, args.speedup_output)
+
+
+if __name__ == "__main__":
+    main()