关于「令牌寿命:暴露窗口,损失上界」,《单兵零信任母典》给出的定量结论是什么?

结论:p42 后半段,取突发 20,每小时 60 次。寿命一小时,最坏 80 次,平均 50 次。寿命六小时,最坏 380 次,平均 195 次。寿命一天,最坏 1,460 次,平均 666 次。寿命一周,最坏 10,100 次,平均 2,668 次。寿命一年,最坏 525,620 次,平均 4,304 次。

数据口径:全书只写防守,不写任何攻击、入侵、绕过或规避检测手法;书中主机、域名、令牌全为合成,参数为示意,全部计算由书中程序生成;本书为非个人化经营分析,不构成安全审计或法律意见,任何生产变更须经人工复核放行。 书中以标注框区分已发表之证据、可运行之 Python 计算(20 段程序)、原典与作者之假说。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 10100, 1460, 195, 20, 2668, 380

rotation scope d L* grid L* sqrt cost/yr never-rot
manual broad200 34.6 29.6 274.9 720.0
manual scoped20 239.8 93.6 69.4 72.0
auto broad200 6.1 5.9 58.4 720.0
auto scoped20 20.6 18.7 17.9 72.0
request ceiling while usable, burst 20 + 60 per hour:
lifetime1 h: worst case 80, mean 50
lifetime6 h: worst case 380, mean 195
lifetime24 h: worst case 1,460, mean 666
lifetime168 h: worst case 10,100, mean 2,668
lifetime8,760 h: worst case 525,620, mean 4,304

核心公式

\[W=\min(R,\ L-A),\]
查看完整验算程序
# Chapter 4. p42: best token lifetime and the loss ceiling it buys.
# Yearly cost of one token with lifetime L hours:
#   C(L) = c 8760 / L + lam d E[W](L),
# c = operator hours per rotation, lam = leaks per token-year, d = loss
# units per hour a leaked token stays usable, E[W] from p41 (exponential
# revocation, mean 72 h). For L much shorter than 72 h, E[W] ~ L / 2 and
# the optimum is the square-root rule L* = sqrt(2 c 8760 / (lam d)).
# Ceiling: a token limited to r requests per hour with burst b can make at
# most b + r W requests while usable (token bucket, chapter 5).
# All parameters illustrative; one loss unit = one operator hour.
import math

M, LAM = 72.0, 0.05


def window(L):
    return M * (1.0 - M / L * (1.0 - math.exp(-L / M)))


def cost(L, c, d):
    return math.fsum([c * 8760.0 / L, LAM * d * window(L)])


def best(c, d):
    grid = [math.exp(i / 200.0) for i in range(0, 1821)]
    return min(grid, key=lambda L: cost(L, c, d))


if __name__ == "__main__":
    print("rotation   scope   d    L* grid  L* sqrt  cost/yr  never-rot")
    for c, ctag in ((0.5, "manual"), (0.02, "auto")):
        for d, dtag in ((200.0, "broad"), (20.0, "scoped")):
            L = best(c, d)
            root = math.sqrt(2.0 * c * 8760.0 / (LAM * d))
            print("{:<9} {:<7} {:>4.0f} {:>8.1f} {:>8.1f} {:>8.1f} {:>9.1f}"
                  .format(ctag, dtag, d, L, root, cost(L, c, d),
                          LAM * d * M))
    b, r = 20, 60
    print("request ceiling while usable, burst {} + {} per hour:".format(
        b, r))
    for L in (1.0, 6.0, 24.0, 168.0, 8760.0):
        print("  lifetime {:>6,.0f} h: worst case {:>8,.0f}, mean {:>7,.0f}"
              .format(L, b + r * min(L, 8760.0), b + r * window(L)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《单兵硅基帝国零信任网安架构、源站隧道隐身与反爬虫防勒索母典》第 4 章「令牌寿命:暴露窗口,损失上界」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-336 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026336,
  author    = {Fan, Yafu},
  title     = {Zero-Trust Cybersecurity for a One-Commander Silicon Empire: Origin Cloaking, Secret Hygiene and Anti-Scraper Defense},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论