Satellites: Evolution & Applications

• Satellites are artificial spacecraft orbiting Earth that support telecommunications, navigation, weather monitoring, and scientific observation.
• They employ diverse architectures—from bent-pipe to regenerative payload systems—to enhance data relay, onboard processing, and network efficiency.
• Current research focuses on sustainability, addressing challenges such as orbital debris, spectrum interference, and integrated space governance.

Searching arXiv for recent, relevant satellite papers to ground the article in current research. I’ll look up broad satellite systems work spanning communications, sustainability, and astronomy impacts. A satellite is an artificial spacecraft placed in orbit to provide communications, positioning, navigation, and timing, weather monitoring, Earth observation, astronomy, and increasingly onboard computation and networking. Since Sputnik 1 in October 1957 marked the beginning of the Space Age, satellites have evolved from isolated platforms into multi-orbit, software-defined, and often networked systems that underpin global data relay, sensing, and scientific observation (Blake, 2022). Contemporary research treats satellites not only as communication relays or sensors, but also as elements of large constellations, edge-computing nodes, and participants in space–air–ground integrated networks; at the same time, their proliferation has made debris, spectrum coexistence, and radio-astronomy interference central engineering and policy constraints (Kodheli et al., 2020).

1. Historical development and orbital regimes

The modern satellite era began with Sputnik 1, whose three-month mission established the feasibility of Earth orbit. Vanguard 1, launched in 1958 and derelict since 1964, remains the oldest artificial object still in orbit because its elliptical medium Earth orbit experiences negligible drag (Blake, 2022). From that starting point, satellite systems diversified into communications, meteorology, navigation, scientific observation, and broadband constellations, while orbital debris accumulated through explosions, collisions, and anti-satellite tests.

Orbit selection governs visibility, latency, path loss, Doppler, lifetime, and operational complexity. LEO is favored when low path loss and low latency are decisive; MEO trades higher delay for fewer spacecraft; GEO provides fixed coverage over the equator but at much larger propagation delay and path loss. In current system design, these regimes are often combined in hybrid architectures rather than treated as mutually exclusive (Kodheli et al., 2020).

Regime	Representative altitude/range	Characteristic properties
LEO	~700–1500 km; period 2–4 hours	Low latency and path loss; high Doppler and frequent handovers (Shutao, 2024)
MEO	~8,000–20,000 km; period 4–12 hours	Intermediate latency and constellation size (Shutao, 2024)
GEO/GSO	~35,786 km altitude over the equator	Fixed footprint; large path length and RTT (Blake, 2022)

The recent New Space transition adds VLEO ~100–450 km, mega-constellations, cloud-managed ground stations, software-defined payloads, and inter-satellite links. A central consequence is that satellite system design is now increasingly constrained by dynamic topology management and coexistence rather than by single-platform link closure alone (Kodheli et al., 2020).

2. Functional classes and platform architectures

Satellites underpin several distinct but increasingly overlapping service classes. Communications satellites relay television, internet, telephony, and broadband; GNSS constellations supply positioning, navigation, and timing; meteorological satellites enable near real-time monitoring of large-scale phenomena; Sun-synchronous and other Earth-observation satellites support environmental monitoring and climate models; and space telescopes bypass atmospheric seeing and absorption for astronomy (Blake, 2022). Satellite IoT extends this landscape toward small, power-constrained terminals and bursty telemetry, with examples including Orbcomm with 36 LEO satellites, Iridium with 66 LEO satellites, Globalstar with 48 LEO satellites, Argo at ~850 km, and planned Chinese constellations such as Hongyan with 60 LEO satellites and Xingyun with 80 LEO satellites (Shutao, 2024).

Architecturally, the primary distinction remains between transparent bent-pipe payloads and regenerative payloads. Bent-pipe systems amplify and frequency-convert traffic while deferring baseband processing to the ground. Regenerative payloads demodulate, decode, switch, and re-encode onboard, reducing feeder-link coupling and enabling onboard routing. Digital Transparent Processors occupy an intermediate position by digitizing the waveform for channelization, beamforming, predistortion, and carrier/power control without full regeneration (Kodheli et al., 2020). Inter-satellite links, whether RF or optical, extend this architecture into dynamic meshes that support cooperative routing, data offloading, and feeder-link reduction.

This architectural spectrum is now broadening toward direct-to-device and disaggregated non-terrestrial networks. One techno-economic analysis of LEO direct-to-device evaluates bent-pipe, regenerative, and 3D Open RAN architectures for unmodified 3GPP-compliant handsets, dimensioned to guarantee ≥ 10 Mbps/user globally. Under the modeled assumptions, net revenue exceeds total cost of ownership with ROI = 13.23%, and monthly cost per subscriber is lowest for the 3D Open RAN option in all three analyzed systems (Aijaz et al., 6 Oct 2025). A plausible implication is that architecture selection is becoming as much an economic optimization problem as a link-budget problem.

3. Communications, networking, and system-level performance

Satellite communication performance is still anchored in the standard radiometric and link-budget relations. The free-space path loss is written as

$$L_{\mathrm{fs}}(\mathrm{dB}) = 32.45 + 20\log_{10}(f_{\mathrm{MHz}}) + 20\log_{10}(d_{\mathrm{km}})$$

and Friis transmission as

$$P_r = P_t G_t G_r \left(\frac{\lambda}{4\pi R}\right)^2.$$

With

$$\mathrm{SNR} = \frac{P_r}{kTB}, \qquad C = B\log_2(1+\mathrm{SNR}),$$

system design becomes a trade among orbit height, antenna gain, bandwidth, and interference (Kodheli et al., 2020). In LEO NTN, Doppler is a first-order effect: the survey reports approximately 350 kHz @14 GHz, 500 kHz @20 GHz, and 750 kHz @30 GHz, which directly constrains synchronization, waveform design, and handover control (Kodheli et al., 2020).

At the network level, inter-satellite communication for small satellites requires coordinated physical, MAC, and routing design. The small-satellite ISL survey emphasizes S-band and X-band RF links, compact patch or array antennas, and MAC strategies ranging from CSMA/CA to TDMA, CDMA, LDMA, and TDMA/CDMA hybrids. For clustered systems, the surveyed TDMA/CDMA hybrid reaches throughput of ≈ 95%, whereas CSMA/CA degrades under higher load, especially in dense clusters (Radhakrishnan et al., 2016). For intermittently connected architectures, DTN with Bundle Protocol and Contact Graph Routing is the natural complement.

Several papers refine this general picture for dense LEO broadband. Beam management under dynamic topology, beam hopping, and satellite–terrestrial spectrum sharing can be cast as a long-term queue stabilization problem. In one Lyapunov-based design, the proposed control reduces the average data queue length of beam cells by over 50% with affordable handover frequency, and in the simulated configuration the maximum network capacity reaches ≈98.51% of a ≈8.08 Gbps upper bound without interference (Sun et al., 2024). In hybrid terrestrial/LEO rural connectivity, stochastic-geometry analysis shows that there exists an optimal constellation density and that average rate can exceed the terrestrial-only baseline while association probability shifts between tiers according to base-station density and satellite density (Salem et al., 2023, Park et al., 2021).

Ground-segment scaling has become equally important. Traditional dishes offer high gain but slow mechanical steering. A distributed phased-array approach, ArrayLink, coherently combines 16 panels, each 32×32 elements, over a ~1.414 km × 1 km aperture. The reported combined gain is ≈48.1 dBi, which is within 1–2 dB of a 1.47 m reflector, while supporting up to four [parallel](https://www.emergentmind.com/topics/additive-parallel-correction) streams at ranges of hundreds of kilometers and two streams beyond ~2000 km (Vennam et al., 12 Aug 2025). For optical feeder links, a HAPS-enabled DTN architecture uses stratospheric relays at ~20 km; over 90 days, LEO–HAPS contacts are 25% more frequent and 25% longer on average than LEO–OGS contacts, materially improving delivery ratio under weather-limited optical downlinks (Fettes et al., 2024).

Empirical mobility measurements show that user-side dynamics are often dominated by occlusion rather than by motion itself. The Starlink Robot, a mobile platform with a Starlink Mini, reports RTT typically ranges 20–40 ms in unobstructed areas, with ≈15 s handovers visible as step changes; at ≈0.8 m/s and ≈2.0 m/s, RTT remains concentrated in the 35–45 ms range, whereas tree-lined paths induce frequent spikes to 40–100 ms (Liu et al., 24 Jun 2025). This suggests that sky visibility, not pedestrian-speed kinematics, is the primary impairment for current mobile LEO broadband.

4. Onboard computing, software maintenance, and autonomous operations

Modern satellites are no longer restricted to forwarding or sensing. In LEO, spacecraft increasingly carry COTS processors such as the Raspberry Pi 4B, memory, and sometimes accelerators to run onboard applications including image encoding, object and ship detection, feature tracking, attitude determination, and onboard data compression (Wen et al., 16 Sep 2025). This shift from remote sensor to orbital compute node reduces downlink load and enables time-sensitive decisions to be taken in orbit.

Software maintenance under these conditions is constrained by intermittent connectivity and low uplink rates. One measured operating point in the update literature is 4–6 contacts per day, about 10 minutes each, with 200 kbps as a representative S-band uplink (Wen et al., 16 Sep 2025). SateLight addresses this by containerizing heterogeneous applications, generating content-aware differential updates on the ground, reconstructing them onboard with fine-grained edits, and recovering via layer-based rollback. Across 10 representative satellite applications, it reduces transmission latency by up to 91.18% (average 56.54%) compared to the best currently available baseline and achieves 100% update correctness; a real in-orbit case on the BUPT-2 satellite demonstrates operational viability (Wen et al., 16 Sep 2025).

Payload security has likewise moved onboard. A real-time video encryption scheme based on two 1D chaotic maps was deployed on a LEO satellite payload for the first time and tested on a Raspberry Pi 4B in the Tiansuan Constellation. For 30 s 360p@20 fps video, end-to-end encryption completed in 2.24–2.47 s; for 10 s 720p@20 fps, in 2.66–2.82 s; power remained stably around ~2.6 W; and the key space is 2^128 (Qiu et al., 20 Mar 2025). The same work reports matching FPGA simulation on Xilinx Zynq xc020clg484, which is significant because integer-only shift–rotate–multiply logic is materially easier to harden than heavier floating-point designs.

Constellation autonomy also extends to integrity monitoring. An ephemeris-free clock phase jump detector based on dual one-way inter-satellite ranges models the constellation as a graph and monitors the fourth singular value of the GCEDM on 5-clique subgraphs. The method is validated on both a 31-satellite GPS constellation and a 17-satellite lunar hybrid constellation, and it does not require prior knowledge of satellite positions or clock biases (Iiyama et al., 2 May 2025). A plausible implication is that future lunar and heterogeneous multi-operator constellations can support onboard timing integrity without centralized ephemeris alignment.

5. Satellites in scientific observation and radio-frequency interference

Satellites extend astronomy by enabling instruments above the atmosphere, but they also increasingly constrain ground-based observation. Space telescopes avoid atmospheric seeing and absorption, and satellites support transient astronomy, including searches associated with gravitational-wave counterparts (Blake, 2022). At the same time, active constellations have become pervasive sources of radio-frequency interference for radio astronomy across much of the spectrum (Peel et al., 15 Apr 2025).

The current operational core of large constellations is in the 10–20 GHz range, but filings and deployments now extend downward and upward: 1.190–1.995 GHz direct-to-cell downlink, 37.5–42.5 GHz user and gateway channels, and proposals in the 120–170 GHz range are explicitly discussed (Peel et al., 15 Apr 2025). Out-of-band leakage is a problem even in protected bands, notably the 10.6–10.7 GHz Radio Astronomy Service allocation adjacent to permitted constellation downlinks. At lower frequencies, unintended electromagnetic radiation has been measured between 110–188 MHz, at 137.5 and 159.4 MHz, and with spectral power flux densities of 15–1300 Jy between 56–66 MHz and 2–100 Jy near 120 and 161 MHz for second-generation Starlink satellites (Peel et al., 15 Apr 2025).

At observatory level, the severity is already clear. QUIJOTE reports geostationary satellites in 10–14 GHz that can appear “as bright as the Sun”, with newer maps showing widespread LEO detections across the sky. LOFAR detects broadband and narrowband Starlink UEMR with flux densities from ~0.1–10 Jy and 10–500 Jy. An SKA-Low prototype images trains of satellites reaching ~10^6 Jy/beam. At millimeter wavelengths, SPT-3G detects thermal emission from satellites at 95, 150, and 220 GHz, consistent with ~300 K blackbody radiation, on timescales of tens of milliseconds (Peel et al., 15 Apr 2025).

These effects enter radio telescopes through both main-beam and sidelobe coupling. In radiometric terms,

$$T_A = \frac{S A_{\mathrm{eff}}}{2k}, \qquad \Delta T = \frac{T_{\mathrm{sys}}}{\sqrt{B\tau}},$$

so any satellite-induced increment in flux density $S$ raises $T_A$, increases $T_{\mathrm{sys}}$, and degrades sensitivity. For interferometers, RFI may be correlated, producing strong visibilities and fringes, or uncorrelated, raising system temperature without forming fringes. The paper emphasizes that calibration, dynamic range, polarimetry, imaging fidelity, and time-domain searches are all affected (Peel et al., 15 Apr 2025).

Mitigation has therefore become a combined technical and regulatory problem. Demonstrated measures include boresight avoidance between the Green Bank Telescope and Starlink, EarthCARE radar shutoff over radio observatories, extension of radio quiet zones, global station databases via the IAU Centre for the Protection of the Dark and Quiet Sky, and prospective protection of the far side of the Moon as a radio-quiet site. The far lunar hemisphere offers natural shielding from terrestrial transmissions, but only if lunar-orbiting satellites are themselves constrained (Peel et al., 15 Apr 2025).

6. Sustainability, debris, and long-term governance

Satellite growth has made the orbital debris environment a primary systems issue rather than a peripheral safety concern. By the end of 2019, confirmed on-orbit fragmentation events totaled 561. The first recorded on-orbit fragmentation occurred in 1961, when a Thor–Ablestar upper stage explosion created roughly 300 trackable fragments. More recent benchmarks include China’s 2007 Fengyun-1C ASAT event, the 2009 Iridium 33–Cosmos 2251 collision with more than 2000 trackable fragments, India’s 2019 ASAT, and Russia’s 2021 ASAT, which created debris intersecting the ISS altitude (Blake, 2022).

Tracking remains incomplete. The US Space Surveillance Network can reliably track objects down to roughly 5–10 cm in LEO and ~1 m at higher altitudes, yet even millimeter-scale debris is dangerous in the hypervelocity LEO regime (Blake, 2022). Models disagree by an order of magnitude in the 1–3 mm size range. ESA’s Sentinel-1A suffered a sub-centimeter impact in 2016, causing permanent partial power loss. Several LEO bands are already assessed as being on the cusp of collisional cascade, and relative velocities often exceed 10 km/s, so ~1 cm debris can be mission-fatal (Blake, 2022).

Operationally, conjunction risk is no longer rare. In 2020, the uncontrolled derelicts IRAS and GGSE-4 passed within ~11 m. Risk modeling is commonly written as

$R = n\,\sigma\,v_{\text{rel}},$

with screening-interval probability

$$P_c \approx 1 - e^{-\int n\sigma v_{\text{rel}}\,dt}.$$

At low altitudes, atmospheric drag governs reentry through

$$a_{\text{drag}} = -\tfrac{1}{2}\,\frac{C_D A}{m}\,\rho\,v^2,$$

and the ballistic coefficient

$$P_r = P_t G_t G_r \left(\frac{\lambda}{4\pi R}\right)^2.$$0

sets decay rate; the 1989 solar maximum temporarily reduced the cataloged LEO population by increasing atmospheric density and drag (Blake, 2022).

Mitigation is therefore multi-layered. The IADC guidelines adopted by UNCOPUOS provide the current foundation. Key engineering measures are passivation, minimizing released material, post-mission disposal, and active debris removal. In LEO, the 25-year rule remains the common post-mission benchmark; in GEO, operators move satellites to graveyard orbits (Blake, 2022). Modeling cited in the sustainability review finds that removing five high-priority targets per year together with ~90% [PMD](https://www.emergentmind.com/topics/patch-mean-decoupling-pmd) compliance could suppress runaway growth under 2003–2011 traffic assumptions, although current launch rates imply that higher removal rates may now be necessary (Blake, 2022).

The long-term governance problem is broader than debris alone. Mega-constellations intensify conjunction assessment, ground-station demand, astronomical contamination, and spectrum conflict. The same trend is pushing research toward improved SSA/SDA, cislunar tracking, hybrid air–space–ground architectures, and more autonomous onboard systems (Blake, 2022, Kodheli et al., 2020). This suggests that the defining feature of the contemporary satellite era is not merely the number of spacecraft in orbit, but the need to manage satellites as an interdependent infrastructure spanning communications, science, software, and the sustainability of near-Earth space.