Hongmeng Project: Early Universe & OS Innovation

• The Hongmeng Project is a dual-initiative effort: one strategy targets early-universe cosmology via a lunar interferometer and the other develops a multi-device, TypeScript-based open-source OS.
• The cosmological mission employs a ten-satellite array to measure the 21 cm signal, achieving enhanced sensitivity to dark matter annihilation, decay, and primordial black holes.
• The HarmonyOS segment features a layered architecture with a pluggable kernel and the ArkTS runtime, enabling secure, scalable applications across diverse device ecosystems.

The Hongmeng Project refers, depending on domain, to two major research thrusts: (1) the Hongmeng mission, a proposed lunar-orbiting global 21 cm cosmological experiment pioneering constraints on particle dark matter and primordial black holes; and (2) the Hongmeng/HarmonyOS mobile operating system, a TypeScript-derived, multi-device open-source software ecosystem with unique structural and analysis challenges. Both initiatives are characterized by distinctive technical innovation and research trajectories within their respective fields.

1. Origins and Strategic Motivation

The term “Hongmeng Project” encompasses both an ambitious scientific mission targeting early-universe cosmology and an operating system (Hongmeng, OpenHarmony, HarmonyOS) designed for cross-device deployment.

• Cosmological Mission: The Hongmeng mission is designed to deploy a lunar-orbiting, multi-satellite interferometer for measuring the global 21 cm brightness temperature of the IGM at redshifts $z \sim 10$–$30$. The primary scientific motivation is to probe exotic energy injections—specifically dark matter (DM) annihilation and decay, and Hawking evaporation of primordial black holes (PBHs)—which are imprinted in small departures from the standard 21 cm spectrum at cosmic dawn, beyond the reach of CMB and $\gamma$-ray constraints (Zhao et al., 2024).
• Mobile Software Ecosystem: The Hongmeng/HarmonyOS project originated from the donation of the HarmonyOS source code by Huawei to the OpenAtom Foundation in June 2021, leading to industry-scale open development under the OpenHarmony governance structure. The operating system targets the “1 + 8 + N” market strategy—with one smartphone hub, eight device classes, and extensible embedded deployments—and aims to secure one-third of the Chinese OS market. Design goals include pluggable kernel abstraction, cross-device capability, and a new application runtime based on ArkTS, a TypeScript-derived language (Li et al., 2023).

2. Architecture and Technical Foundations

Hongmeng Mission (Cosmology)

• Instrument Design: Ten-satellite lunar-orbiting interferometer array, covering 0.1–120 MHz (mapping $z \sim 10$–$10^3$). Spectral resolution is 1 MHz per channel across $\sim 100$ channels, with an angular resolution $\theta_{\rm fg} \simeq 5^\circ$ and sky coverage fraction $f_{\rm sky} \simeq 0.8$.
• Systematics Control: Primary foreground is modeled as Galactic synchrotron emission $$T_{\rm fg}(\nu)=16.3 \times 10^6{\rm K}(\nu/2~{\rm MHz})^{-2.53}$$. Lunar far side is exploited for its radio-quiet environment and thermal stability, critical for precision (Zhao et al., 2024).

Hongmeng/HarmonyOS

• Layered Stack: Four canonical layers—Kernel (Linux or LiteOS, unified via Kernel Abstraction Layer), System Service (security, hardware abstraction, core daemons), Framework (ArkTS APIs, ability management, ArkUI), and Application (ArkTS-based .hap bundles).
• ArkTS Runtime: ArkTS supersedes legacy Java stacks, providing single-threaded asynchronous semantics, advanced type features (optionals, decorators), and integration with the Stage-based ability management and ArkUI declarative UI engine.
• Security Model: Adopts capability-granular permissions and mandatory access control at the service layer, coupling these with an API-centric programming interface (Li et al., 2023).

3. Scientific and Engineering Achievements

Cosmological Reach

• Dark Matter: Hongmeng achieves forecasted 1$\sigma$ sensitivity to thermal annihilation cross section $\langle\sigma v\rangle \lesssim$ few $\times 10^{-28}$ cm$^3$ s$^{-1}$ for 10 GeV DM with 1000 h integration and foreground residual $\epsilon_0=10^{-4}$.
• Dark Matter Decay: Sensitivity to decay lifetimes $\tau \gtrsim 10^{28}$ s for $m_\chi \lesssim 1$ GeV, outperforming CMB (Planck), $\gamma$-ray, and Voyager-Cosmic-ray bounds by two orders of magnitude.
• Primordial Black Holes: Constrains $f_{\rm PBH} \lesssim 10^{-7}$–$10^{-8}$ for $M_{\rm PBH} \sim 10^{16}$ g, complementing and surpassing existing constraints from galactic $\gamma$-rays and neutrino backgrounds (Zhao et al., 2024).

Platform and Ecosystem

• Development Metrics: OpenHarmony 4.0 contains 10,435 ArkTS APIs, with the count growing every release cycle. In the open-source repository, >600 framework contributors and 3,804 candidate repos have been harvested, with 174 confirmed OpenHarmony apps and 96 ArkTS libraries distributed via the ohpm package registry.
• Toolchain and Governance: Maintains its own IDE (DevEco Studio), build, test, and debugging pipelines, layered module governance (Operating Committee, Technical Steering Committee, domain-specific TSGs) (Li et al., 2023).

4. Analysis, Benchmarking, and Technical Methodologies

Hongmeng Mission: Signal and Constraint Modeling

• Observables: The differential sky-averaged 21 cm brightness temperature relative to the CMB is modeled by

$$\delta T_{21}(z)=23\,{\rm mK} \times x_{\rm HI}(z)\Biggl(\frac{0.15}{\Omega_m} \Biggr)^{1/2} \Biggl( \frac{\Omega_b h^2}{0.02} \Biggr) \Biggl(\frac{1+z}{10}\Biggr)^{1/2}[1-T_{\rm CMB}(z)/T_S(z)]$$

The spin temperature $T_S$ is computed via Wouthuysen–Field effect.

• Energy Injection Models: DM annihilation, decay, and PBH evaporation inject non-standard energy, modifying the IGM thermal/ionization evolution.
• Statistical Forecasts: An analytic Fisher-matrix framework is applied to forecast multi-parameter sensitivities, revealing that astrophysical and exotic (DM, PBH) parameters exhibit weak degeneracy, enabling robust joint fitting (Zhao et al., 2024).

Hongmeng API Synthesis and LLM Methods

• Code Generation Challenge: LLMs yield poor performance on ArkTS due to limited exposure, with pass@1 scores for off-the-shelf models ranging from 0–18% on the OHBen 300-task benchmark.
• API Knowledge Graph (API KG): APIKG4SYN systematically constructs an API knowledge graph from HarmonyOS SDK documentation, extracts single/multi-API seeds, and generates synthetic question–code pairs by prompting LLMs and employing uncertainty-driven Monte Carlo Tree Search for multi-API unfamiliarity maximization.
• Empirical Results: Fine-tuned Qwen2.5-Coder-7B achieves 25.00% pass@1, outperforming GPT-4o (17.59%). Multi-API data are essential; removing them degrades results sharply (25.00% → 10.19%) (Liu et al., 29 Nov 2025).

Model	w/o SFT	APIKG4SYN SFT	GPT-4o (zero-shot)
Qwen2.5-Coder-7B	8.33%	25.00%	17.59%
Mistral-7B	4.63%	14.81%	—
Deepseek-R1-7B	0.00%	9.26%	—

5. Comparison with Peer Platforms and Missions

Aspect	OpenHarmony	Android	iOS
Kernel	Linux or LiteOS + KAL	Linux (Bionic)	XNU
App Language	ArkTS (TypeScript-based)	Java/Kotlin, NDK	Swift, Objective-C
API Count	≈10,435	>30,000	~20,000
Security Model	Capability + Mandatory AC	Permission + SELinux	Sandbox, Entitlements
Multi-Device	Built-in ("1+8+N")	Separate UI kits	Separate OSes by device
OSS Community	~600 contributors	~10,000 (framework)	Closed core; some open libs

Hongmeng’s cosmological mission uniquely fills the $z\sim10$–$30$ low-frequency window inaccessible to ground-based telescopes due to ionospheric cutoff, extending DM/PBH sensitivity by up to two orders of magnitude versus CMB/$\gamma$-ray/microlensing bounds. HarmonyOS diverges via its kernel abstraction (KAL), ArkTS language, and unified ability-lifecycle model, requiring dedicated static/dynamic analysis and compliance frameworks (Zhao et al., 2024, Li et al., 2023).

6. Open Challenges and Research Directions

Cosmology

• Systematics: Foreground residual control at $\epsilon_0\lesssim10^{-4}$–$10^{-3}$ is critical; algorithmic advances (e.g., AI-based cleaning) are required.
• Integration Time: $t_{\rm int}\sim1000$ h per patch is essential for deep thermal noise suppression.
• Extension: Sensitivity may be extended to earlier "dark ages" ($z\gtrsim30$); multi-epoch analysis remains an open field (Zhao et al., 2024).

Software Engineering

• ArkTS Analysis: No open-source static analysis or IR builders exist for ArkTS; porting from Java/Kotlin or Swift/Objective-C tools is non-trivial due to language and packaging divergence.
• Dynamic Instrumentation: Absence of ArkTS runtime hooks impedes dynamic/hybrid security and resilience testing.
• Security/Ecosystem Tools: No GDPR, privacy, or fairness vetting pipelines tailored to OpenHarmony; black-market malware detectors are lacking.
• LLM Code Assistance: Further advances require (1) mining open-source ArkTS projects for code augmentation, (2) generation of API misuse exemplars, and (3) live retrieval of API KG context during code generation to reduce error rates (Li et al., 2023, Liu et al., 29 Nov 2025).

7. Implications, Recommendations, and Prospects

The Hongmeng Project, in both its cosmological and software contexts, exemplifies a confluence of system-level innovation, methodological rigor, and domain-specific research priorities.

• Cosmological Impact: By leveraging lunar-based radio-quiet conditions, extended integration, and precision modeling, the Hongmeng mission offers a direct probe of exotic early-universe physics, with the capacity to close major parameter-space windows and force new constraints on DM and PBH theories. This suggests that similar mission designs may be adopted internationally for next-generation 21 cm cosmology (Zhao et al., 2024).
• Software Development and Ecosystem: The ArkTS-centric stack under OpenHarmony advances multi-device, secure, and privacy-respecting software deployment but presents nontrivial migration and analysis barriers relative to incumbent mobile OSes. Targeted construction of ArkTS static/dynamic analysis tools, ecosystem vetting frameworks, and LLM-centric code assistant pipelines is essential. A plausible implication is that cross-platform adaptation (e.g., ReactNative/Flutter support) will depend on the maturation of interoperability and widget mapping layers (Li et al., 2023, Liu et al., 29 Nov 2025).

Research communities are advised to engage in foundational toolchain construction—especially ArkTS parsing, GUI/modeling, and permission/security analysis—while augmenting code generation and compliance checking via API knowledge graph methods and uncertainty-driven synthetic data generation. These steps are necessary prerequisites for the Hongmeng Project to achieve parity with, and potentially exceed, established platforms in both scientific and technological domains.