Introduction

Optical Inter-Satellite Links (OISL) use free-space laser communication to transmit data between satellites at speeds approaching 200 Gbps. Unlike radio frequency links, optical systems achieve higher bandwidth through shorter wavelengths and tighter beam collimation. The ArkSpace Exocortex Constellation proposes using OISL for neural spike train transmission between orbital nodes.

This technology sits at TRL 6-7 (demonstrated in operational environments). Commercial systems exist, but adapting protocols for neural data represents novel engineering work at TRL 2.

Physical Principles

OISL operates in the near-infrared spectrum, typically at 1550 nm wavelength (C-band telecom standard). This wavelength matches fiber optic communication infrastructure, allowing reuse of mature ground-based components.

Laser Beam Propagation

A laser terminal transmits a collimated beam with divergence measured in microradians. At 2,000 km distance (typical inter-satellite range in LEO), a 10 microradian beam divergence creates a spot size of approximately 20 meters in diameter. The receiving telescope collects photons from this spot.

Link Budget Fundamentals: Free space path loss follows the equation: L = (4πd/λ)², where d is distance and λ is wavelength. At 1550 nm and 2,000 km range, path loss reaches approximately 281 dB. High transmit power and large aperture telescopes compensate for this loss.

Pointing Requirements

Satellite-to-satellite links at LEO velocities (~7.5 km/s relative motion) require precision pointing. Star trackers provide coarse attitude knowledge (±10 arcsec), while fine pointing assemblies achieve ±0.1 arcsec accuracy. Acquisition time for initial beam lock typically takes <10 seconds.

The OISL terminal must track the target satellite’s predicted trajectory using orbital propagation data (Two-Line Elements or higher-precision ephemeris). Kalman filters integrate attitude sensor data with orbital mechanics models to maintain beam alignment.

Commercial OISL Systems

TESAT Laser Communication Terminals

TESAT Spacecom manufactures flight-proven OISL terminals used in the European Data Relay System (EDRS). Operational performance:

• Data rate: 1.8 Gbps (demonstrated in orbit since 2016)
• Range: Up to 45,000 km (GEO to LEO links)
• Mass: ~35 kg per terminal
• Power: ~120W during active transmission
• Wavelength: 1064 nm

TESAT’s system prioritizes reliability over maximum bandwidth. The conservative 1.8 Gbps rate ensures robust operation under varying atmospheric conditions (for ground links) and thermal fluctuations.

Mynaric Optical Communication Terminals

Mynaric produces commercial OISL hardware targeting LEO constellations:

• Claimed data rate: Up to 100 Gbps (product specifications, not independently verified in operational deployments)
• Form factor: Compact design for CubeSat integration
• Mass: <15 kg (terminal head)
• Status: TRL 6, ground demonstrations completed

Mynaric’s higher bandwidth claims reflect advances in photodetector sensitivity and coherent optical modulation schemes. Field testing at scale remains limited compared to TESAT’s operational heritage.

Starlink Gen2 OISL Claims

SpaceX’s FCC filings for Starlink Gen2 satellites claim optical links exceeding 100 Gbps. Independent verification of operational performance does not exist in public literature. Proprietary system details remain undisclosed.

If achieved, these data rates would represent a 55× improvement over EDRS performance. The technical path likely involves wavelength-division multiplexing (WDM), coherent detection, and advanced error correction codes.

Bandwidth Requirements for Neural Data

Neural spike trains encoded in Address-Event Representation (AER) consume bandwidth based on firing rates:

Single Neuron:

• 8 bytes per spike (32-bit neuron address + 32-bit timestamp)
• At 100 Hz average firing rate: 800 bytes/sec = 6.4 Kbps

100 Million Neuron Payload:

• Idle state (10 Hz average): ~6.4 Gbps
• Active state (100 Hz average): ~64 Gbps
• Burst activity (1000 Hz peak): ~640 Gbps

Peak burst activity exceeds even theoretical 200 Gbps OISL capacity. The arkspace-core architecture addresses this through generative models that transmit prediction errors rather than raw spike streams, achieving 90%+ bandwidth reduction. With compression, sustained transmission fits within 1-10 Gbps for typical neural activity patterns.

Latency Analysis

OISL latency has two components:

Propagation Delay (Physics-Limited)

Speed of light in vacuum: 299,792 km/s

• 550 km (ground to satellite): 1.83 ms one-way
• 2,000 km (inter-satellite, typical): 6.7 ms one-way
• 5,400 km (inter-satellite, maximum): 18 ms one-way

These delays are fundamental limits. No technology can transmit faster than light speed.

Protocol Overhead

• Packet framing: 0.1-0.5 ms
• Error correction (Reed-Solomon decoding): 0.2-0.8 ms
• OISL acquisition/tracking (after initial lock): <0.1 ms jitter
• Queue delays: Variable, 0.5-2 ms depending on traffic load

Total Round-Trip Time (RTT) Budget: For a single OISL hop between two satellites at 2,000 km separation:

• Propagation: 6.7 ms × 2 = 13.4 ms
• Protocol overhead: ~1-2 ms
• Total: 15-20 ms RTT

This aligns with the ArkSpace specification of <20 ms RTT for inter-satellite links.

Atmospheric Limitations

OISL operates effectively in vacuum (space-to-space links). Ground-to-satellite optical links face atmospheric challenges:

• Cloud cover: Complete signal blockage
• Atmospheric turbulence: Beam scintillation and wander
• Rain: Scattering losses (less severe than RF Ka-band, but not negligible)

The Exocortex Constellation design uses optical links exclusively for satellite-to-satellite communication. Ground links employ Ka-band RF (26.5-40 GHz) to maintain connectivity during adverse weather.

Technology Gaps for Neural Applications

While OISL hardware exists (TRL 6-7), several adaptations are required for neural data:

Protocol Optimization (TRL 2)

Standard CCSDS Space Packet Protocol targets file transfers and telemetry. Neural spike streams have different characteristics:

• Continuous real-time data (not file-based)
• Latency-sensitive (cannot buffer for retransmission)
• High redundancy (some spike loss may be tolerable)
• Prediction error encoding (non-standard compression)

The arkspace-core repository proposes a custom OISL neural protocol with 64-byte headers, AES-256-GCM authentication tags, and priority-based packet scheduling. This protocol exists as specification documents only, no implementation exists.

Power Budget Constraints (TRL 3-4)

Current OISL terminals consume 30-120W during active transmission. Operating two terminals simultaneously (for mesh connectivity) adds 60-240W to the satellite power budget. Combined with neuromorphic payload (50-100W), total power draw reaches 110-340W.

A 12U CubeSat with 400W solar arrays can sustain this load in sunlight, but eclipse periods (up to 35 minutes per 96-minute orbit) drain battery reserves. Power management strategies remain under analysis.

Radiation Tolerance (TRL 4-6)

Laser diodes and photodetectors experience degradation under radiation exposure. Commercial OISL terminals use radiation-tolerant components, but total ionizing dose (TID) ratings vary:

• EDRS (GEO environment): Qualified for 15 krad over 15-year mission
• LEO requirements (550 km): 50 krad over 5-year mission

Component selection and shielding strategies require detailed analysis for long-duration LEO deployment.

Error Correction Requirements

Bit error rates (BER) on optical links typically range from 10⁻⁹ to 10⁻¹² (uncoded). Neural data transmission may tolerate higher error rates if the neuromorphic processor can interpolate missing spikes.

The proposed Reed-Solomon (223, 255) code corrects up to 16 byte errors per 255-byte block. This adds 14% overhead but reduces BER to below 10⁻¹⁵.

Forward Error Correction (FEC) encoding introduces latency. Hardware accelerators (FPGA-based) can reduce encoding/decoding time to <1 ms, but this requires custom development (TRL 3-4).

Comparison with RF Links

Traditional satellite communication uses radio frequency links (S-band, X-band, Ka-band). OISL offers distinct advantages and tradeoffs:

Parameter	OISL (1550 nm)	Ka-band RF (30 GHz)
Bandwidth	60-200 Gbps	100 Mbps - 1 Gbps
Antenna size	10-30 cm telescope	30-100 cm dish
Pointing accuracy	±0.1 arcsec	±0.1°
Weather sensitivity	Immune (space-space)	Rain fade (ground)
Spectrum regulation	Unlicensed	ITU coordination required
Maturity (TRL)	6-7	8-9

OISL provides 60-200× higher bandwidth but requires precision pointing two orders of magnitude tighter than RF systems. For inter-satellite neural data transmission, OISL’s bandwidth advantage outweighs the pointing complexity.

Cost Analysis

OISL terminal procurement costs (approximate, based on vendor quotes and public contracts):

• TESAT heritage terminals: $1.5M - $2M per unit
• Mynaric commercial terminals: $800K - $1.5M per unit (quoted pricing, volume dependent)
• Custom development (adapted for neural protocols): +$500K - $1M NRE (non-recurring engineering)

Each satellite requires two OISL terminals for mesh connectivity. Total communication hardware cost per node: $1.6M - $4M.

Ground-based testing infrastructure adds:

• Optical terminal testbed: $300K - $500K
• Atmospheric chamber (thermal-vacuum): $200K - $400K
• Bit error rate test equipment: $100K - $200K

Current Deployment Status

As of January 2026:

• EDRS operational: 1.8 Gbps links between GEO relay satellites and LEO observation satellites
• Starlink Gen2: Claims of 100+ Gbps, no independent verification
• NASA LCRD (Laser Communications Relay Demonstration): Demonstrated 1.2 Gbps from geosynchronous orbit
• JAXA LUCAS: Tested OISL on ISS (link speeds not publicly disclosed)

No operational constellation uses OISL exclusively for inter-satellite communication. Starlink represents the largest deployment at scale, but technical details remain proprietary.

Integration with Neuromorphic Payloads

OISL terminals interface with neuromorphic processors via PCIe Gen4 x8 (126 Gbps bidirectional). Data flow:

1. Neuromorphic processor generates spike events (8 bytes per spike)
2. DMA controller transfers spike buffers to OISL interface card
3. Protocol stack adds headers, encryption (AES-256-GCM), FEC
4. Laser modulator encodes data stream
5. Fine pointing assembly directs beam to target satellite

Latency from neuromorphic processor output to photon emission: <1 ms (target specification, not demonstrated).

Path to Implementation

Adapting OISL for neural data requires:

Phase 1 (2026-2027): Protocol specification, simulation of neural traffic patterns, latency modeling.

Phase 2 (2028-2029): Ground-based hardware testbed with commercial OISL terminals, protocol stack implementation, integration with neuromorphic processor simulators.

Phase 3 (2030+): On-orbit demonstration mission with single OISL link, validation of latency and bandwidth under real space conditions.

This timeline assumes neuromorphic processors become available (currently TRL 3-4). OISL hardware availability is not the bottleneck. Protocol development and system integration represent the critical path.

Summary of Findings

Optical Inter-Satellite Links provide the bandwidth foundation for distributed neural computing in orbit. Commercial hardware from TESAT and Mynaric demonstrates TRL 6-7 maturity, with operational systems achieving 1.8 Gbps and claimed development of 100+ Gbps terminals.

Adapting OISL for neural spike train transmission requires protocol development (TRL 2), power optimization (TRL 3-4), and integration with neuromorphic processors (TRL 2-3). The physics of laser communication imposes fundamental latency limits (~7 ms per 2,000 km), but this remains well within the <20 ms specification required for real-time neural processing.

Technology gaps exist primarily in software and integration, not in OISL hardware. Closing these gaps requires focused engineering effort but does not depend on breakthroughs in optical physics.

---

Official Sources

1. ArkSpace OISL Protocol Specification: arkspace-core/docs/protocols/oisl-neural-protocol.md
2. TESAT Laser Communication Terminals: TESAT Spacecom GmbH Product Specifications.
3. Mynaric Optical Terminals: Mynaric AG Product Datasheets, 2025.
4. EDRS System Performance: European Space Agency, “European Data Relay System: Five Years of Operations,” 2021.
5. Starlink Gen2 FCC Filings: SpaceX, FCC Application SAT-MOD-20200417-00037, April 2020.
6. NASA LCRD: NASA Goddard Space Flight Center, Laser Communications Relay Demonstration Mission Updates, 2023.
7. Optical Link Budget Analysis: Hemmati, H. (2006). “Deep Space Optical Communications,” Wiley-Interscience.
8. Free-Space Optics Fundamentals: Kaushal, H., & Kaddoum, G. (2017). “Optical Communication in Space: Challenges and Mitigation Techniques,” IEEE Communications Surveys & Tutorials, 19(1), 57-96.
9. CCSDS Space Packet Protocol: Consultative Committee for Space Data Systems, CCSDS 133.0-B-2, 2020.