Digest · Link physics · read after Chapter 2
導讀 · 鏈路物理 · 第 2 章後閱讀
Recent On-Orbit Results and ARQ Performance Analysis for the TBIRD 200-Gbps MissionTBIRD 200 Gbps 任務的近期在軌結果與 ARQ 效能分析
TBIRD demonstrates a high-rate optical downlink using commercial telecom transceivers. Selective-repeat ARQ recovers data lost during atmospheric fades, while oscillator pre-compensation keeps Doppler within the receiver window. Its ground telescope and downlink geometry define the scope of this flight result.TBIRD 使用商用電信收發器示範高速光學下行，選擇性重傳 ARQ 復原大氣衰落造成的資料遺失，本地振盪器預補償則讓多普勒偏移維持在接收窗口內。地面望遠鏡與下行幾何共同界定這項飛行結果的適用範圍。
If you read nothing else
只讀這一段也可以
TBIRD is the one peer-reviewed data point in this primer for a 200 Gbps free-space optical link in orbit. The paper limits its model to one terminal-to-ground link, excluding constellation, node-delay and setup-delay models.TBIRD 是本教材中唯一一個在軌 200 Gbps 自由空間光學鏈路的同儕審查資料點。論文模型範圍排除星座、節點延遲及建立延遲。
| Measure or comparison參數或比較項目 | Values, conditions and evidence數值、條件與證據 |
|---|---|
| Evaluation conditions | NASA's 6U PTD-3 CubeSat carried a 3U payload in a sun-synchronous orbit descending from 530 km toward about 450 km. Commercial telecom transceivers supplied two 100 Gbps DP-QPSK coherent channels. The ground receiver was the 1 m JPL Optical Communications Telescope Laboratory telescope. Its adaptive optics coupled the beam into fiber over ranges of about 600–1,400 km. [37, §I–III] |
| Evaluation conditions | The record pass delivered 4.8 TB in about five minutes. The second campaign delivered over 33 TB across 22 passes, with zero residual errors. At 200 Gbps, the payload drew 100 W and held pointing below 10 µrad RMS. The link ran at 150–200 Gbps whenever the satellite was above about 30° elevation [37, Figs. 4–8] . |
| ARQ and Doppler | Error-free delivery through atmospheric fades came from a selective-repeat ARQ with 1,752 virtual channels that cost under 1% overhead and reached 85–95% efficiency. Doppler caused an instructive failure. The COTS receiver tolerated frequency offsets of about −17 to +21 ppm. Ephemeris-based oscillator pre-shifting by about 1.5 GHz recovered 30% of the data on one pass. [37, §IV, Figs. 9–10] |
| Evaluation conditions | It is a single terminal talking to a single large ground aperture. This evidence anchors Chapter 2's link budget and Chapter 13's downlink discussion. Starlink laser performance lies outside its scope. |
| 星座配置 | 搭載在 NASA 6U PTD-3 立方衛星上的 3U 酬載，在從 530 km 降向約 450 km 的太陽同步軌道上，攜帶兩個以商用電信收發器組成的 100 Gbps DP-QPSK 相干通道，把光束指向 JPL 光通訊望遠鏡實驗室的 1 m 望遠鏡（其自適應光學把光束耦合進光纖），距離約 600–1,400 km[37, §I–III]。 |
| 實測效能 | 紀錄過頂在約五分鐘內送下 4.8 TB，第二次活動的 22 次過頂共送下 33 TB 以上且殘餘錯誤為零，酬載在 200 Gbps 模式下耗電 100 W、指向誤差保持在 10 µrad RMS 以下，而只要衛星仰角高於約 30°，鏈路就以 150–200 Gbps 運作[37, Figs. 4–8]。 |
| 頻率 | 穿越大氣衰落仍能零錯誤遞送，來自一個有 1,752 個虛擬通道、額外負擔低於 1%、效率達 85–95% 的選擇性重傳 ARQ，而這次活動最有教育意義的失敗是多普勒：COTS 接收器只能容忍約 −17 到 +21 ppm 的頻率偏移，所以依星曆把本地振盪器預先偏移約 1.5 GHz，在某次過頂中救回了 30% 的資料[37, §IV, Figs. 9–10]。 |
| 評估條件 | 它是單一終端對單一大型地面孔徑的對話，而這正是它能錨定第 2 章鏈路預算與第 13 章下行問題、Starlink 雷射性能則位於其證據範圍之外的原因。 |
1. Where this paper sits1. 這篇論文的位置
Every other digest in this group reasons about laser links that are assumed to exist. This one reports a laser link that flew. It studies optical link performance through elevation angle, received power and oscillator offset. Networking papers instead tend to study hop count and topology.
For the primer it does three jobs. The NTRS copy carries an "In Review" marking. The primer records that marking and treats the ICSOS conference paper as peer-reviewed. The study covers one satellite, one terminal and one ground station.
Constellation snapshots, node-delay models and setup-delay models lie outside its scope. Each pass does begin with an acquisition, but the fact base leaves TBIRD acquisition time unspecified.
這篇報告實際在軌運作的雷射鏈路，為本組以假設鏈路為主的導讀提供實證。它屬於光學文獻，分析仰角、接收功率與振盪器偏移，網路文獻則多分析跳數與拓樸。對本教材而言它做了三件事。這是一篇會議論文，其 NTRS 版本帶有「審查中」標記，所以本教材基於 ICSOS 的會議身分把它視為同儕審查，同時註明該標記。
星座版本位於這項單衛星研究的範圍之外，因為只有一顆衛星、一個終端、一個地面站，因此節點延遲假設與建立延遲模型亦位於範圍之外。每次過頂確實從一次擷取開始，但事實庫將 TBIRD 擷取時間保留為待補。
| Measure or comparison參數或比較項目 | Values, conditions and evidence數值、條件與證據 |
|---|---|
| Evaluation conditions | It gives the receiver sensitivity that Chapter 2's link budget is checked against. This is a FEC threshold of about −41 dBm for a 100 Gbps channel [37, §II]. |
| Link or data rate | It provides the only peer-reviewed satellite demonstration of a 200 Gbps class optical rate. SpaceX's 100 Gbps sustained and 200 Gbps peak per-laser figures remain company statements. |
| Evaluation conditions | And it gives the per-pass data volume, 4.8 TB, that Chapter 13 uses to size the front end of an orbital data center. |
| 評估條件 | 它給出第 2 章鏈路預算用來核對的接收器靈敏度，即 100 Gbps 通道約 −41 dBm 的 FEC 門檻[37, §II]。 |
| 鏈路速率證據 | 它給出唯一一個衛星 200 Gbps 等級光學速率的同儕審查示範，相比之下 SpaceX 宣稱的每個雷射 100 Gbps 持續、200 Gbps 峰值是公司說法。 |
| 評估條件 | 它也給出第 13 章用來估算軌道資料中心前端規模的每次過頂資料量 4.8 TB。 |
2. The problem, measured2. 問題，用量測說話
TBIRD addresses the Earth-observation downlink bottleneck described in Chapter 8. A small satellite can generate more data than a radio pass returns. The conventional larger antenna exceeds the available volume of a CubeSat. The mission's measurement is therefore the volume returned per pass and the conditions under which the link holds. The third is the receiver's tolerance to Doppler, measured the hard way.TBIRD 處理的問題是第 8 章描述的地球觀測下行之牆：一顆小衛星產生的資料遠多於一次無線電過頂能送回的量，而加大無線電天線的傳統答案體積超出立方衛星的可用空間。所以這個任務的量測是每次過頂送回的資料量，以及鏈路能維持的條件。第三個是接收器對多普勒的容忍度，從實際鏈路失敗中量得。
| Measure or comparison參數或比較項目 | Values, conditions and evidence數值、條件與證據 |
|---|---|
| Power budget | During the second campaign, the satellite descended from about 530 km toward about 450 km. It linked with the 1 m OCTL telescope over roughly 600–1,400 km. Rates reached 150–200 Gbps above about 30° elevation. The full 200 Gbps mode drew 100 W of payload power. [37, §I–III, Figs. 4–8] |
| Evaluation conditions | The record pass on 17 May 2023 returned 4.8 TB in about five minutes, roughly a terabyte a minute. The average over the 22 passes was about 1.5 TB, and the campaign total exceeded 33 TB with zero residual errors. |
| Evaluation conditions | The second measured quantity is pointing. The paper reports this as below 10 µrad RMS with the payload feeding corrections to the spacecraft bus at 10 Hz [37, §I–II]. |
| Frequency | On 24 April 2023, the link dropped near −16 ppm frequency offset. This established a COTS acquisition window of about −17 to +21 ppm with compensation disabled. The team then pre-shifted the oscillator by about +1.5 GHz, around 8 ppm at 1.5 µm. This centered the expected −3 GHz peak shift within the window. The 17 May pass increased from roughly 3.3 TB to the 4.8 TB record. The authors attribute the 30% recovery entirely to Doppler compensation. [37, §IV, Figs. 9–10] |
| 星座配置 | 在第二次活動中，位於約 530 km 並降向約 450 km 的衛星，以約 600–1,400 km 的距離與 1 m 的 OCTL 望遠鏡建鏈，只要仰角超過約 30° 就維持 150–200 Gbps，完整的 200 Gbps 模式耗用 100 W 酬載功率[37, §I–III, Figs. 4–8]。 |
| 評估條件 | 2023 年 5 月 17 日的紀錄過頂在約五分鐘內送回 4.8 TB，大致每分鐘一個 TB，而 22 次過頂的平均約為 1.5 TB，活動總量超過 33 TB 且殘餘錯誤為零。 |
| 指向精度 | 第二個量測量是指向，論文報告為低於 10 µrad RMS，酬載以 10 Hz 把修正量回饋給衛星本體（摘要與 Secs. I–II）。 |
| 頻率 | 在 2023 年 4 月 24 日的過頂中，頻率偏移到達約 −16 ppm 時鏈路中斷，這確立了 COTS 收發器在補償停用下約 −17 到 +21 ppm 的可用窗口，而當團隊依星曆把本地振盪器預先偏移約 +1.5 GHz（在 1.5 µm 約 8 ppm），讓預期約 −3 GHz 的峰值偏移落在窗口內之後，5 月 17 日的過頂從約 3.3 TB 升到 4.8 TB 的紀錄，這 30% 的回收完全歸因於多普勒補償[37, §IV, Figs. 9–10]。 |
| Question問題 | Formula公式 | Result結果 |
|---|---|---|
| Average rate over the record pass紀錄過頂的平均速率 | \( 4.8\times10^{12}\times 8 / 300\ \text{s} \) | ≈ 128 Gbpscomputed. "about five minutes" taken as 300 s. The instantaneous rate was 150–200 Gbps only above ~30° elevation計算值。「約五分鐘」取 300 秒。瞬時速率只在仰角 ~30° 以上達 150–200 Gbps |
| Average volume per pass in the second campaign第二次活動每次過頂的平均資料量 | \( 33\ \text{TB}/22 \) | ≈ 1.5 TBcomputed. Matches the paper's stated average [37, §III]計算值。與論文所述平均一致[37, §III] |
| Carrier frequency at 1,550 nm and the Doppler window in GHz1,550 nm 的載波頻率與以 GHz 表示的多普勒窗口 | \( f=c/\lambda;\; \Delta f = f\times\text{ppm} \) | 193.4 THz. −3.3 to +4.1 GHzcomputed from the −17 to +21 ppm window [37, §IV]由 −17 到 +21 ppm 的窗口計算[37, §IV] |
| Range rate behind the ~3 GHz peak shift約 3 GHz 峰值偏移背後的距離變化率 | \( v_r = \Delta f\,\lambda \) | ≈ 4.65 km/scomputed. 0.645 GHz per km/s at 1,550 nm. The pre-shift of 1.5 GHz corresponds to 2.3 km/s計算值。1,550 nm 下每 km/s 為 0.645 GHz。1.5 GHz 的預偏移對應 2.3 km/s |
| Doppler of a counter-moving LEO pair against this window反向運行 LEO 衛星對的多普勒與此窗口的比較 | \( 15\ \text{km/s}\times 0.645\ \text{GHz/(km/s)} \) | ≈ 9.7 GHz ≈ 33 ppmcomputed. Outside the −17 to +21 ppm window, so crossing-plane ISLs need LO pre-compensation計算值。超出 −17 到 +21 ppm 窗口，所以交叉軌道面 ISL 需要本地振盪器預補償 |
| Margin above the −41 dBm threshold in Chapter 2's 10 cm, 1 W, 1,550 nm example第 2 章 10 cm、1 W、1,550 nm 範例高於 −41 dBm 門檻的餘裕 | \( P_R(2{,}000\ \text{km})=-28\ \text{dBm} \) | 13 dB at 2,000 km, ~5 dB at 5,000 kmcomputed in Chapter 2 from the Friis form. The threshold is TBIRD's [37, §II]第 2 章以 Friis 形式計算。門檻為 TBIRD 的值[37, §II] |
3. The design3. 設計
The design idea is to buy the modem and build only what space requires. Turbulence causes millisecond fades in the optical downlink. A fiber telecom modem treats each fade as a burst of lost frames. TBIRD therefore adds selective-repeat automatic repeat request.
On the ground, a 1 m telescope with adaptive optics collects the beam and couples it into single-mode fiber. This permits a commercial receiver. Space-to-space links require a separate receive design. A coherent receiver must lock its local oscillator to the incoming carrier.
The telecom DSP acquires signals over only tens of ppm, requiring Doppler compensation. Thus, the oscillator is pre-shifted from the predicted range rate before each pass.
設計理念是買現成的調變解調器，只打造太空所必需的部分。穿越湍流的光學下行會一次衰落數毫秒，而為光纖設計的電信調變解調器把每次衰落當成一串遺失的幀，所以 TBIRD 把資料流包進選擇性重傳自動重複請求裡。地面端是設計的另一半，也是星對星鏈路須另行設計的那一半：一座帶自適應光學的 1 m 望遠鏡收集光束並耦合進單模光纖，讓同樣的商用接收器可以在地面使用。多普勒修正補齊全貌：相干接收器必須把本地振盪器鎖到入射載波，而電信 DSP 只能在數十 ppm 內拉入，所以每次過頂前依預測的距離變化率預先偏移振盪器。
| Measure or comparison參數或比較項目 | Values, conditions and evidence數值、條件與證據 |
|---|---|
| Evaluation conditions | The two 100 Gbps channels use dual-polarization QPSK coherent telecom transceivers widely deployed in terrestrial fiber networks. TBIRD therefore inherits their forward error correction and −41 dBm class sensitivity. [37, §II] |
| Evaluation conditions | A fine-pointing system sends 10 Hz corrections to the bus and keeps the beam within 10 µrad RMS. A separate link layer handles atmospheric fades. |
| Evaluation conditions | Data are buffered on the satellite, the ground reports which frames failed. Only those frames are resent, with the scheme spread over 1,752 virtual channels so that the retransmission bookkeeping keeps pace with 200 Gbps. |
| Evaluation conditions | The measured cost is under 1% overhead and an ARQ efficiency of 85–95%. The measured benefit is a campaign total above 33 TB with zero residual errors [37, §III–IV, Figs. 4–8]. |
| 評估條件 | 兩個 100 Gbps 通道是地面光纖網路以千計使用的雙偏振 QPSK 相干收發器，所以酬載繼承了它們的前向錯誤更正與 −41 dBm 等級的靈敏度，將開發工作集中於太空所需的支援功能[37, §II]。 |
| 精指向系統 | 團隊在它們周圍打造的是一套精指向系統，以 10 Hz 向衛星本體送出修正、把光束保持在 10 µrad RMS 以內，以及一個能撐過大氣的鏈路層。 |
| 星座配置 | 資料在衛星上緩衝，地面回報哪些幀失敗，只重送那些幀，整個機制分散在 1,752 個虛擬通道上，讓重傳的簿記跟得上 200 Gbps。 |
| 評估條件 | 量到的成本是低於 1% 的額外負擔與 85–95% 的 ARQ 效率，量到的效益是活動總量超過 33 TB 且殘餘錯誤為零[37, §III–IV, Figs. 4–8]。 |
Original figure note原始圖說出處
Drawn for this primer from Schieler et al., ICSOS 2023 (NTRS 20230014010), Secs. II–IV and Figs. 4–10.4. Evaluation setup4. 實驗設定
The evaluation is an on-orbit measurement campaign rather than a simulation. This makes it the most trustworthy source in this group and also the narrowest. The link is a downlink. Thus, the receive aperture is a metre across and sits under an atmosphere that the ARQ is designed to defeat. The study has one satellite, with filing snapshots, node delay and routing outside its scope.
There is one optical terminal rather than three or four. Each pass is a scheduled contact. The reviewed material leaves acquisition time unspecified, so setup delay lies outside the model. The paper leaves the space terminal's transmit power and aperture diameter unspecified.
An earlier SPIE paper gives these parameters and remains outside this source review. Thus, the link budget requires additional source inputs.
評估採用在軌量測活動，這使它成為本組最可信的來源，同時也是最窄的。鏈路是下行，所以接收孔徑有一公尺寬，而且位於 ARQ 所要克服的大氣之下。研究只有一顆衛星，申報版本、節點延遲及路由皆位於範圍之外。只有一個光學終端，其他星座的三或四終端配置則屬另一情境。
建立延遲位於模型範圍之外，因為每次過頂是排定的接觸，而在事實庫驗證的資料中論文將擷取時間保留為待補。論文也將太空終端發射功率與孔徑直徑保留為待補，這些存在於一篇本次來源範圍之外的較早 SPIE 論文中，所以單靠這份文件須加入額外輸入才能閉合鏈路預算。
| Measure or comparison參數或比較項目 | Values, conditions and evidence數值、條件與證據 |
|---|---|
| Ground infrastructure | The second campaign covers 22 passes from PTD-3, a 6U spacecraft in a sun-synchronous orbit between about 530 and 450 km. Links span about 600–1,400 km to the 1 m OCTL telescope with adaptive optics. [37, §I–III] |
| Link or data rate | Rates are reported as instantaneous channel rates versus elevation and as per-pass volumes. The Doppler findings come from two specific passes, 24 April and 17 May 2023. |
| 星座配置 | 一顆 6U 衛星 PTD-3，在約 530 到 450 km 之間的太陽同步軌道上，與一個地面站（帶自適應光學的 1 m OCTL 望遠鏡）通訊，第二次活動共 22 次過頂、距離約 600–1,400 km[37, §I–III]。 |
| 鏈路或資料速率 | 速率以相對仰角的瞬時通道速率與每次過頂資料量報告，多普勒的發現則來自 2023 年 4 月 24 日與 5 月 17 日兩次特定過頂。 |
5. What was measured5. 量到了什麼
| Claim宣稱 | Setup and caveat設定與附註 |
|---|---|
| 200 Gbps optical downlink from a CubeSat using two 100 Gbps DP-QPSK COTS channels.以兩個 100 Gbps DP-QPSK COTS 通道從立方衛星達成 200 Gbps 光學下行。 | LEO to a 1 m adaptive-optics telescope at 600–1,400 km [37, §I–III]. The measurement covers a satellite-to-ground link. The receive aperture is 10× the diameter of a 10 cm ISL terminal.LEO 對 600–1,400 km 的 1 m 自適應光學望遠鏡[37, §I–III]。範圍限於星對地鏈路。接收孔徑直徑是 10 cm ISL 終端的 10 倍。 |
| 4.8 TB in one ~5 min pass. >33 TB error-free over 22 passes. ~1.5 TB average per pass.一次約 5 分鐘過頂 4.8 TB。22 次過頂 >33 TB 零錯誤。每次過頂平均約 1.5 TB。 | Figs. 4–8. The record pass used Doppler compensation, the average includes passes before it and lower-elevation geometry.Figs. 4–8。紀錄過頂使用了多普勒補償，平均值包含之前的過頂與較低仰角的幾何。 |
| 150–200 Gbps sustained above ~30° elevation. 100 W payload power in 200 Gbps mode.仰角約 30° 以上維持 150–200 Gbps。200 Gbps 模式酬載功率 100 W。 | Sec. III. Below 30° the atmospheric path and range reduce the rate, so the pass-average is nearer 128 Gbps [computed].Sec. III。30° 以下大氣路徑與距離降低速率，所以過頂平均接近 128 Gbps［計算值］。 |
| Pointing error <10 µrad RMS with 10 Hz feedback to the bus.指向誤差 <10 µrad RMS，以 10 Hz 回饋衛星本體。 | Abstract, Secs. I–II. The only demonstrated pointing figure in the fact base, used in Chapter 2 in place of unpublished Starlink values.摘要、Secs. I–II。事實庫中唯一的實證指向數字，第 2 章用它提供可查證的實證參考，Starlink 對應數值仍屬專有資訊。 |
| Selective-repeat ARQ: 1,752 virtual channels, <1% overhead, 85–95% efficiency.選擇性重傳 ARQ：1,752 個虛擬通道、額外負擔 <1%、效率 85–95%。 | Secs. III–IV. Requires on-board buffering and a feedback channel whose latency the fact base leaves unrecorded.Secs. III–IV。需要機上緩衝與一條回饋通道，其延遲須另查來源。 |
| COTS receiver Doppler window ≈ −17 to +21 ppm. LO pre-shift of +1.5 GHz recovered 30% of a pass (3.3 → 4.8 TB).COTS 接收器多普勒窗口約 −17 到 +21 ppm。本地振盪器預偏移 +1.5 GHz 救回一次過頂 30% 的資料（3.3 → 4.8 TB）。 | Sec. IV, Figs. 9–10. Threshold observed at about −16 ppm on 24 April 2023. A counter-moving LEO pair would sit at ~33 ppm [computed], outside the window.Sec. IV、Figs. 9–10。2023 年 4 月 24 日觀察到約 −16 ppm 的門檻。反向運行的 LEO 衛星對約在 33 ppm［計算值］，超出窗口。 |
6. Limits and open questions6. 限制與待解的問題
- The result describes an optical downlink. The 1 m receive telescope has a hundred times the collecting area of a 10 cm space terminal. This is 20 dB of the budget. Its adaptive optics do work that a space receiver treats through a separate optical design. The 200 Gbps therefore demonstrates the transceivers and the pointing, with a 2,000–5,000 km satellite-to-satellite link requiring a separate budget. The primer uses TBIRD as a receiver-sensitivity reference. SpaceX claims 100 Gbps sustained and 200 Gbps peak per laser. Journalists relay these company statements (fact base Part C, item 14 company).此結果描述光學下行鏈路。1 m 的接收望遠鏡有 10 cm 太空終端一百倍的集光面積，相當於預算中的 20 dB，而它的自適應光學做的工作太空接收器須以另一套光學設計處理。因此 200 Gbps 示範的是收發器與指向，2,000–5,000 km 的星對星鏈路則須另行計算，本教材把它當作靈敏度錨點，SpaceX 每個雷射 100 Gbps 持續、200 Gbps 峰值宣稱的證據則仍限於透過記者轉述的公司說法（事實庫 Part C 第 14 項 company）。
- The link budget requires additional source inputs. Transmit power and space-aperture diameter are unspecified in the ICSOS text the fact base verified. Thus, the −41 dBm threshold is the only end of the budget available here. Chapter 2 closes the loop with a worked 1 W, 10 cm example that is the primer's computation, with TBIRD's transmitter parameters treated separately.閉合鏈路預算需要其他來源的參數。發射功率與太空孔徑直徑仍待從事實庫驗證的 ICSOS 文字以外查證，所以 −41 dBm 的門檻是此處唯一可用的預算端。第 2 章用 1 W、10 cm 的範例閉合迴路，那是本教材的計算，TBIRD 的發射參數則須另行核對。
- Headline versus average. The 4.8 TB record was one pass after Doppler compensation. The campaign average was about 1.5 TB and the pass-average rate about 128 Gbps [computed]. A budget of 4.8 TB per pass uses the best case. A continuous 200 Gbps budget uses a rate achieved only above 30° elevation.標題數字對平均數字。4.8 TB 的紀錄是多普勒補償後的一次過頂。活動平均約 1.5 TB、過頂平均速率約 128 Gbps［計算值］，所以以每次過頂 4.8 TB 預算的前端設計用的是最佳情況，以持續 200 Gbps 預算的設計用的是只在 30° 仰角以上才成立的速率。
- The Doppler window is a warning for crossing-plane ISLs. The −17 to +21 ppm tolerance equals about 3–4 GHz at 1,550 nm. Counter-moving LEO satellites approaching at 15 km/s produce about 9.7 GHz or 33 ppm [computed]. Coherent COTS-based ISLs in that geometry therefore require ephemeris-based oscillator pre-compensation, as TBIRD used. This is one more reason the temporary links of the companion digests are hard. Same-direction adjacent-plane links in a 53° shell see only about 0.17 GHz [computed in Chapter 2].多普勒窗口是對交叉軌道面 ISL 的警告。−17 到 +21 ppm 的容忍度在 1,550 nm 約為 3–4 GHz，而兩顆以 15 km/s 接近的 LEO 衛星產生約 9.7 GHz 或 33 ppm［計算值］，所以任何反向運行衛星之間以 COTS 為基礎的相干 ISL 都必須像 TBIRD 一樣依星曆預補償振盪器，這是姊妹導讀中暫時鏈路困難的又一個理由。53° 軌道殼中同向相鄰軌道面鏈路只看到約 0.17 GHz［第 2 章計算］。
- ARQ needs a return path and a buffer whose size requires another source. Selective repeat requires ground feedback within the pass. Feedback latency and buffer depth require additional source records. For a store-and-forward data return this is immaterial. For the interactive traffic of Chapters 5–6 it would set the retransmission RTT.ARQ 需要一條回程與一個容量須另查來源的緩衝區。選擇性重傳預設地面能在過頂內告訴衛星哪些幀失敗，而回饋通道延遲與緩衝深度都須另查來源。對儲存轉送式的資料回傳這些參數對總資料量影響較小。對第 5–6 章的互動式流量它會決定重傳 RTT。
- One site, one satellite, a reviewed copy. All passes used the OCTL telescope, so weather and site statistics are absent. The NTRS copy carries an "In Review" marking. The venue is a refereed conference, but readers citing exact figures should check the final ICSOS proceedings version.一個站、一顆衛星、一份審查中的版本。所有過頂都用 OCTL 望遠鏡，天氣與站點統計須由多站量測補充，而 NTRS 版本帶有「審查中」標記。會議本身有審查，但引用精確數字的讀者應核對 ICSOS 會議論文集的最終版本。
7. How it connects7. 與其他內容的關聯
- Section 2.2 uses the −41 dBm receiver threshold. The worked link budget has a 13 dB margin above it at 2,000 km.第 2.2 節：−41 dBm 門檻是衡量範例鏈路預算在 2,000 km 下 13 dB 餘裕的接收器靈敏度。
- Section 2.3 lists TBIRD in the 200 Gbps row. The table also covers LCRD, EDRS, the SDA standard and company-tagged SpaceX figures.第 2.3 節：TBIRD 是已飛行與規格速率表中的 200 Gbps 那一列，與 LCRD、EDRS、SDA 標準及標為公司說法的 SpaceX 數字並列。
- Section 2.4: the <10 µrad pointing figure and the ppm Doppler window are the demonstrated values behind the pointing and Doppler discussion.第 2.4 節：<10 µrad 的指向數字與 ppm 級多普勒窗口是指向與多普勒討論背後的實證數值。
- Section 2.6 compares Suncatcher's 100–200 m links. They use coherent DWDM transceivers of the same class, with far-field 1/d² loss outside their operating regime.第 2.6 節：與 Suncatcher 100–200 m 鏈路的對比，後者使用同一類相干 DWDM 收發器，卻完全擺脫 1/d² 損耗。
- Section 8.1 and Section 13.5 discuss the downlink bottleneck. Research direction D sizes an orbital data center front end around 200 Gbps and 4.8 TB per pass.第 8.1 節與第 13.5 節：TBIRD 所攻擊的下行之牆，以及研究方向 D 用來估算軌道資料中心前端規模的 200 Gbps 等級、每次過頂 4.8 TB 的光學下行。
- Sibling digests arxiv23-setup-delay and vtm21-chaudhry explain crossing-plane constraints. The Doppler window reinforces those constraints. The short-range alternative appears in joule26-suncatcher. Radio downlink bottlenecks appear in sigcomm21-l2d2 and asplos20-oec.同組與相關導讀：arxiv23-setup-delay 與 vtm21-chaudhry 說明交叉軌道面鏈路為何困難，多普勒窗口強化了這一點。joule26-suncatcher 是短距離的替代方案。sigcomm21-l2d2 與 asplos20-oec 則是無線電下行瓶頸。
Key numbers關鍵數字
| Number | 數字 | Meaning (setup) | 意義（設定） |
|---|---|---|---|
| 200 Gbps | Two 100 Gbps DP-QPSK coherent channels on COTS transceivers, LEO to ground [37, §II] | 兩個以 COTS 收發器組成的 100 Gbps DP-QPSK 相干通道，LEO 對地[37, §II] | |
| 6U / 3U | PTD-3 spacecraft and TBIRD payload volume. ~530 → ~450 km sun-synchronous orbit [37, §I] | PTD-3 衛星與 TBIRD 酬載體積。約 530 → 約 450 km 太陽同步軌道[37, §I] | |
| 600–1,400 km | Link range to the 1 m JPL OCTL telescope with adaptive optics [37, §III] | 到帶自適應光學的 1 m JPL OCTL 望遠鏡的鏈路距離[37, §III] | |
| 4.8 TB / ~5 min | Record pass, 17 May 2023, after Doppler compensation [37, Figs. 4–8] | 紀錄過頂，2023 年 5 月 17 日，多普勒補償後[37, Figs. 4–8] | |
| >33 TB / 22 passes | Campaign-2 total, error-free. ~1.5 TB average per pass [37, §III] | 第二次活動總量，零錯誤。每次過頂平均約 1.5 TB[37, §III] | |
| 150–200 Gbps above ~30° | Rate versus elevation. Pass-average ≈ 128 Gbps [computed] | 速率對仰角。過頂平均約 128 Gbps［計算值］ | |
| 100 W | Payload power in 200 Gbps mode [37, §II] | 200 Gbps 模式的酬載功率[37, §II] | |
| <10 µrad RMS | Pointing error, with 10 Hz feedback to the bus [37, §I–II] | 指向誤差，以 10 Hz 回饋衛星本體（摘要、Secs. I–II） | |
| 1,752 / <1 % / 85–95 % | ARQ virtual channels, overhead and efficiency [37, §III–IV] | ARQ 虛擬通道數、額外負擔與效率[37, §III–IV] | |
| −17 to +21 ppm | COTS receiver Doppler window with compensation disabled. +1.5 GHz LO pre-shift recovered 30% of a pass [37, §IV, Figs. 9–10] | 補償停用時 COTS 接收器的多普勒窗口。+1.5 GHz 本地振盪器預偏移救回一次過頂 30% 的資料[37, §IV, Figs. 9–10] | |
| ≈ −41 dBm | FEC threshold of the 100 Gbps transceiver, the sensitivity anchor for Chapter 2 [37, §II] | 100 Gbps 收發器的 FEC 門檻，第 2 章的靈敏度錨點[37, §II] |