1: \begin{abstract}

2: A mega-constellation of low-altitude earth orbit (LEO) satellites (SATs) are envisaged to provide a global coverage SAT network in beyond fifth-generation (5G) cellular systems. 

3: LEO SAT networks exhibit extremely long link distances of many users under time-varying SAT network topology. 

4: This makes existing multiple access protocols, such as random access channel (RACH) based cellular protocol designed for fixed terrestrial network topology, ill-suited. 

5: To overcome this issue, in this paper, we propose a novel grant-free random access solution for LEO SAT networks, dubbed \emph{emergent random access channel protocol (eRACH)}. 

6: In stark contrast to existing model-based and standardized protocols, eRACH is a model-free approach that emerges through interaction with the non-stationary network environment, using multi-agent deep reinforcement learning (MADRL).  

7: Furthermore, by exploiting known SAT orbiting patterns, eRACH does not require central coordination or additional communication across users, while training convergence is stabilized through the regular orbiting patterns.

8: Compared to RACH, we show from various simulations that our proposed eRACH yields $54.6$\% higher average network throughput with around two times lower average access delay while achieving $0.989$ Jain's fairness index.

9: 

10: % Lastly, by utilizing locally observable information, we validate that the eRACH naturally emerges from a given local environment while reducing the communication overhead between user terminals, that highlighting the applicability of the proposed LEO SAT-oriented access.

11: \end{abstract}

12: