縮短低軌衛星通訊初始同步訊號搜索延遲之聯合功率與週期最佳化設計

Abstract

在低軌道（LEO）衛星通訊系統中，同步訊號搜索延遲（cell search delay） 是一項關鍵的效能指標，其表現受同步訊號區塊（SSB）傳輸功率與週期性顯著影響。較高的傳輸功率可提升偵測成功率，而較短的週期有助於縮短等待時間，兩者皆能有效降低小區搜尋遲。然而，這兩種手段都會增加衛星有限的電力負擔，因此必須在功率與週期性之間進行權衡。此外，現行 3GPP的SSB聯組（burst）設計主要針對地面行動網路，對於高度動態的 LEO 衛星環境並不適用。為克服此限制，本論文突破傳統的 SSB突發框架，提出一個聯合SSB功率與週期性最佳化架構，以最小化使用者設備（UE）的小區搜尋延遲。針對此非凸問題，我們首先利用黃金比例搜尋（golden-section search）建立資料驅動的均勻初始化，以確保演算法的穩定收斂；隨後提出一種交替最佳化演算法，將原問題分解為凸性的週期性子問題與基於序列二次規劃（SQP）的功率子問題。在36個六邊形小區覆蓋的 LEO衛星足跡下，針對不同遮蔽萊斯衰落（shadowed-Rician）通道條件進行模擬，結果顯示本方法具有顯著優勢。與均勻功率與週期性配置相比，平均延遲降低 48.9%；相較僅進行功率最佳化與僅進行週期性最佳化，分別提升 15.1%與29.3%；與5G SSB聯組基準（2、4、8聯組配置）相比，則實現 18.6–47.6%的效能增益。這些結果證明了基於 LEO波束跳頻特性的自適應小區層級 SSB資源配置策略的有效性，成功填補了現有研究中的關鍵空白。

Cell search delay, a critical performance metric in Low Earth Orbit (LEO) satellite communication systems, is significantly impacted by Synchronization Signal Block (SSB) transmit power and periodicity. Higher transmit power improves detection success rates while shorter periodicity reduces waiting time, both helping to reduce cell search delay. However, both approaches strain the limited onboard power budget, necessitating careful tradeoffs between power and periodicity. Moreover, current 3GPP SSB burst designs are primarily suited for terrestrial networks and prove inadequate for LEO satellite networks with their highly dynamic environments. Therefore, this thesis breaks from the conventional SSB burst framework and proposes a joint SSB power and periodicity optimization framework to minimize UE cell search delay. To solve this non-convex problem, we first determine a data-driven uniform initialization via golden-section search to ensure robust convergence. Then, a novel alternating optimization algorithm decomposes the original problem into convex periodicity subproblems and sequential quadratic programming (SQP)-based power subproblems. Extensive simulations over 36 hexagonal-cell LEO footprints with varying shadowed-Rician channel severities demonstrate superior performance. The proposed method yields a 48.9% average delay reduction compared to uniform power and periodicity allocation, a 15.1% improvement over power-only optimization, a 29.3% improvement over periodicity-only optimization, and 18.6–47.6% gains against 5G SSB burst baselines with 2, 4, and 8 bursts. These results highlight the efficacy of adaptive cell-level SSB resource allocation tailored to LEO beam-hopping dynamics, addressing a critical gap in prior research.