SDN-Orchestrated Dual-Path 5G/SATCOM Maritime Communications for Carrier Strike Groups Avinash Srinivasan∗ , Dalibor Španjević† , ‡ , Kevin H. Nguyen§ , ‡ , Bannon Ireton§ , ‡ , and Christopher B. Landis∗
arXiv:2609.05729v1 [cs.NI] 4 Sep 2026
∗ Cyber Science Department, United States Naval Academy, Annapolis, MD. Email: {srinivas, clandis}@usna.edu † Carey Business School, Johns Hopkins University, Baltimore, MD. Email: [email protected] § United States Navy, Pensacola, FL. Email: {kevinnguyen2791, banto.ire20}@gmail.com
Abstract—carrier strike group (CSG) communications must sustain mission traffic across heterogeneous links whose quality varies with distance, weather, fading, and intermittent outages. We present an integrated Software-Defined Networking (SDN)/Software-Defined Radio (SDR) framework that unifies 5G and satellite communications (SATCOM) under a single control plane for dual-path failover in CSG communications. The framework is evaluated via stochastic simulation across four operational regions: Norfolk, the Norwegian Sea, the Philippine Sea, and the North Pacific storm track. The framework couples a centralized SDN controller with a Rician-faded 5G channel under Adaptive Modulation and Coding (AMC), a Ku-band SATCOM channel with ITU-R P.618/P.838 rain attenuation, and Gilbert–Elliott burst-availability processes on both links. The controller routes each packet by traffic class and current link state. Experiments use 60 independent trials of 50,000 packets per parameter setting with common random numbers, paired confidence intervals, and convergence diagnostics across fault, fading, and region sweeps. For the Norfolk weather baseline, SDN routing reaches 93.09% reliability versus 36.22% (5G-only) and 89.87% (SATCOM-only), with paired gains of +56.87 and +3.22 percentage points. The advantage grows under joint low-Rician-K/high-burst-fault stress, and AMC improves 5G robustness as Signal-to-Interference-plusNoise Ratio degrades. SDN remains the highest-reliability mode across all four regions, indicating that cross-layer SDNcontrolled failover improves maritime resilience without relying on deterministic link abstractions. Index Terms—Carrier Strike Group, maritime communications, software-defined networking, software-defined radio, 5G, SATCOM, adaptive modulation and coding
I. Introduction Modern carrier strike groups (CSGs) operate in environments where continuous, reliable, and resilient communications are critical to mission success [1]. A CSG ‡ Work completed at the U.S. Naval Academy. The views expressed in this document are those of the authors and do not reflect the official policy or position of the U.S. Naval Academy, Department of the Navy, the Department of War, or the U.S. Government. This is the authors’ accepted manuscript of a paper accepted for publication in the 2026 IEEE Military Communications Conference (MILCOM 2026), 12–16 October 2026. © 2026 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
contains an aircraft carrier, guided-missile escorts, logistics support, and other mission-dependent assets that must coordinate across wide geographic areas and dynamic threat environments. In this work, we abstract that force into a carrier with escort ships, a satellite relay, and a shore node so that routing behavior can be studied without tying the analysis to a specific country’s CSG deployment. These ships rely on a mix of satellite communications (SATCOM) links to exchange sensor data, Command and Control (C2) messages, and logistics information over Internet Protocol (IP) [2]. However, traditional communication architectures are often hardware centric and statically configured, making it difficult to adapt to degraded lineof-sight (LoS), adverse weather, jamming, intermittent outages, and fluctuating bandwidth demands [1], [3]. Conventional afloat networks also lack a unified mechanism for prioritizing traffic across heterogeneous links. A tactical air-defense update has a tighter latency and assurance requirement than routine administrative data, yet both may compete for the same physical channels and routing mechanisms. When rain fade, distance-dependent path loss, multipath fading, or bursty link outages degrade a path, the network may not automatically reroute traffic over alternatives such as ship–carrier–ship relays or SATCOM backhaul. This rigidity can waste scarce satellite resources when ships are within LoS, and it can also leave mission traffic exposed when a preferred low-latency path becomes unavailable [2], [3]. Software-Defined Networking (SDN) and SoftwareDefined Radio (SDR) technologies offer a path toward adaptive maritime communications [4]. An SDN separates the control and data planes, enabling a logically centralized controller to observe network state, enforce mission-aware policies, and change forwarding behavior dynamically [5], [6]. An SDR complements this controlplane flexibility at the maritime radio: shipboard radios can sense sea-state-, range-, and weather-driven channel variation, adapt Adaptive Modulation and Coding (AMC) as link quality changes, and expose actionable link-state information to higher-layer routing decisions [3]. Together, SDN and SDR motivate a dual-path cross-layer architecture in which tactical traffic normally prefers low-latency
5G, administrative traffic normally prefers SATCOM, and the controller can fail over to avoid degraded paths. Programmable maritime networks combining 5G and SATCOM under unified SDN/SDR control have been proposed in prior architectural work [1], [4], [6]. However, many simulation studies typically rely on deterministic link abstractions or scenario-level fault gates that do not capture packet-level variability in fading, weather, and burst errors [7], [8]. This paper addresses these gaps by implementing a physically grounded stochastic simulation: a Rician-faded 5G link with AMC, a Ku-band SATCOM link with International Telecommunication Union (ITU)R P.618 [9] rain attenuation, and Gilbert–Elliott bursterror availability for both paths, evaluated under paired stochastic conditions. The simulated topology (Fig. 1) comprises the aircraft carrier (hosting gNodeB and SDN controller), five escort ships, a geostationary (GEO) satellite for ship-shore relay, and a shore command node; details are formalized in Section III-A. The controller classifies each packet as tactical or administrative, observes link state and Physical Layer (PHY) feasibility, and selects between the 5G and SATCOM paths with per-packet failover. Performance is measured using packet delivery reliability, tactical deadline-miss rate, SATCOM offload, latency jitter, and mean latency. The key contributions of this work are: • A stochastic dual-path CSG simulator that captures packet-level variability in fading, weather, and link availability across both 5G and SATCOM paths, enabling realistic operational assessment under stochastic link stress. • A per-packet mission-aware SDN controller with cross-layer use of channel state, AES-256-GCM accounting for tactical traffic on SATCOM, and three routing modes: sdn, static_5g (5G-only), and static_satcom (SATCOM-only) for paired comparison. • A rigorous evaluation protocol providing statistically grounded inter-mode comparisons via paired-sample inference and convergence diagnostics. • A multi-region evaluation showing how fault rate, Rician K-factor, traffic mix, and weather distributions affect SDN reliability and failover behavior. The remainder of this paper is organized as follows. Section II reviews related work on SDN, SDR, hybrid maritime links, and resilient network simulation. Section III presents the system model and experimental methodology. Section IV reports the simulation results, sensitivity studies, and multi-region validation. Section V concludes the paper with future research directions. II. Related Work Prior maritime SDN/SDR work has built the architectural case for programmable control over heterogeneous shipboard links [4], [6], [10]. In general, a logically centralized controller can manage paths, allocate
Ku-band SATCOM Shore (C2)
Carrier Ship 1 5G
gNodeB + SDN SATCOM failover
Ship 2 Carrier SATCOM pipe
Fig. 1. Simulated CSG: carrier-hosted gNodeB/SDN controller, escorts with 5G primary, SATCOM failover, GEO ship–shore relay, and shore node. Two of five escorts shown.
scarce spectrum, and react to changing link conditions in ways that statically configured naval IP infrastructure cannot [2]. Cross-layer designs that expose SDR channel state to higher-layer routing decisions further extend this control surface to the physical layer [3], [11]. Niknami et al. [1] proposed a unified SDN–SDR-driven cross-layer maritime communications framework that leverages the existing SATCOM infrastructure for resilient communications in dynamic, resource-constrained environments; our presented work implements and quantitatively evaluates that framework under stochastic channel, weather, and burst-availability conditions. These prior contributions are largely architectural; they motivate SDN control for naval communications but stop short of such evaluation. A. SDN for Multi-SATCOM and Cognitive Maritime Links A complementary thread targets the multi-SATCOM aggregation problem. The SDN-SAT lineage models each ship as an SDN switch over several segregated SATCOM terminals and uses Multipath TCP under centralized control for load balancing and throughput optimization [12], [13], with later extensions adding UAV relays for handover continuity in mobile tactical scenarios [14]. Ghafoor and Koo [15] take a different route, layering SDNbased cognitive routing over cognitive-radio maritime links with cluster-head local views. These efforts validate their controllers in Mininet-class emulators or custom simulators with opaque links, leaving the packet-level interaction between PHY-layer impairments and SDN routing decisions outside their scope. B. Simulation and Fault Modeling SDN simulation studies for resilient networks typically validate controller logic, topology effects, and failover behavior under deterministic outages, independent periodic faults, or scenario-level weather gates [7], [8]. Such abstractions can obscure packet-level effects that drive maritime link behavior, including multipath fading, weatherdriven attenuation, and bursty outages. Common random
numbers (CRN) and paired statistical comparisons separate routing-policy effects from random environmental variation [16], but this discipline is rarely combined with physically grounded PHY models in the maritime SDN literature. III. System Model and Methodology This section presents the simulation model: a physicallygrounded, stochastic, dual-path link model coupled to a centralized SDN controller with an adaptive SDR front end. Relative to the deterministic abstractions common in prior maritime SDN studies [4], [7], the model comprises: 1) a Rician-faded 5G channel with AMC; 2) a Ku-band SATCOM channel with ITU-R P.618 rain attenuation driven by per-quarter-hour observed weather; 3) a Gilbert–Elliott [17], [18] two-state link-availability model; and 4) an experimental protocol with CRN, paired hypothesis tests, and run-count and packet-count convergence diagnostics following Law and Kelton [16]. A. Topology, Traffic, and Operational Regions The CSG is modeled as seven surface nodes: one aircraft carrier (co-located gNodeB and SDN controller host), five guided-missile escorts at a mean baseline carrier-to-escort separation of ≈ 20 km (ranging 1.1 to 44 km across the high-, mid-, and low-band 5G envelopes), and one shore command facility. Surface node positions are specified by geodesic coordinates and pairwise inter-node separations are computed via the Haversine formula, d = 2RE arcsin (√ ( sin2
∆ϕ 2
)
( + cos ϕ1 cos ϕ2 sin2
∆λ 2
)) ,
(1)
where RE = 6,371 km is Earth’s mean radius, ϕ is latitude, and λ is longitude. The GEO SATCOM relay is treated as a separate link abstraction characterized by a fixed elevation angle (45◦ ) and an effective rain-layer path length, rather than as a positioned node in the topology. The SATCOM channel is modeled with a fixed slantrange geometry, capturing rain fade, propagation delay, and return-channel overhead. A per-region treatment with elevation-angle-dependent rain attenuation is left to future work. To isolate the effect of regional weather distributions, the same CSG topology and SATCOM geometry are simulated under weather samples drawn from four operational regions: Norfolk, VA, USA (mid-Atlantic baseline), the Norwegian Sea (68◦ N, polar edge), the Philippine Sea (tropical), and the North Pacific storm track (50◦ N, −155◦ W). Differences in metrics across regions thus reflect differences in rain-rate distributions rather than fleet geometry. Traffic is generated as a stream of packets denoted by Np , each tagged tactical or administrative. Tactical
TABLE I AMC table used by the 5G radio. MCS
Min SINR (dB)
Spectral Eff. (b/s/Hz)
2 5 10 16 22
0.5 1.0 2.0 4.0 6.0
BPSK QPSK 16-QAM 64-QAM 256-QAM
packets are short (256 B nominal), latency-sensitive with a representative deadline Dtac = 100 ms (chosen as a lowlatency tactical messaging target), and prefer the lowestlatency feasible link; administrative packets are larger (4 kB) and tolerant of SATCOM transit. The fraction of generated packets tagged tactical, πtac , is a sweep parameter (default 0.5, equal mix). B. Dual-Path PHY Model 1) 5G Link with Rician Fading and AMC: For each packet, the instantaneous 5G Signal-to-Interference-plusNoise Ratio (SINR), γ = S/(I + N ) [19], is drawn from a Rician distribution parameterized by K-factor, where K controls the ratio of deterministic LoS power to diffuse multipath power, and a deterministic path-loss term whose breakpoints determine band selection: highband (d ≤ 1.6 km), mid-band (1.6 < d ≤ 20 km), and lowband (20 < d ≤ 55 km), where 55 km is the 5G coverage cutoff beyond which packets are routed to SATCOM. Conditioned on SINR γ, the radio selects the highest modulation and coding scheme (MCS) whose minimum SINR is met (Table I) [19], [20]. This makes the SDR element of the framework explicit: the PHY senses the channel and adapts. An ablation with AMC disabled fixes the scheme and drops packets whose SINR cannot support it. 2) Ku-band SATCOM with Rain Fade: The SATCOM channel models propagation delay, transmission delay, and Ku-band rain attenuation per Recommendation ITU-R P.618 [9]. For specific attenuation γR (dB/km) at frequency f and rain rate R (mm/h), with frequency-dependent coefficients k(f ), α(f ) tabulated in ITU-R P.838 [21], γR = k(f ) Rα(f ) ,
(2)
with slant-path attenuation A = γR LE (θ, hR ) accumulated along an effective path length LE that depends on elevation θ and rain-cell height hR . The end-to-end SATCOM latency is L 2nh + + nr T r , (3) c Rb where n counts hops through the GEO relay and shore node, L is payload size, Rb ∈ {2, 4, 8} Mbps is the negotiated data rate, nr is the retry count, and Tr a retry penalty. τsat =
TABLE II Gilbert–Elliott link-availability parameters (baseline). Parameter Steady-state bad prob. πB Mean bad-burst length L̄B (packets) Error prob. in good state Error prob. in bad state
5G
SATCOM
0.05 6 (PHY) 1.00
0.02 8 (PHY) 1.00
3) Weather Sampling: Rain rate is treated as a perpacket stochastic input rather than a scenario-level constant. The simulator ingests one year (May 2025–April 2026) of 15-min Open-Meteo observations at the chosen operational region, and for each packet draws an independent and identically distributed (i.i.d.) rain-rate sample R from this empirical distribution. The sampled R enters the SATCOM link budget as ITU-R P.618 rain attenuation, so fade events affect individual packets according to the observed weather distribution of the region rather than its annual mean. The i.i.d. per-packet sampling is a deliberate worst-case stress test of the routing layer: it preserves the marginal rain-rate distribution observed in the region but breaks the temporal correlation of sustained storm episodes, exposing the controller to maximum-variance rain-fade transitions on every packet. Coupling rain rate to the Gilbert–Elliott bad-state process to model storm persistence as a continuous-time Markov channel is left as future work; the current results should be read as an upper bound on per-packet fade variability under realistic marginal weather. C. Link Availability: Gilbert–Elliott Model Independent of fading and rain, each link transitions between good and bad states as a two-state Markov chain [17], [18]. While bad, the link drops packets at a high error probability; while good, error probability is set by the PHY. The chain is parameterized by the steady-state bad probability πB and the mean bad-state burst length L̄B (Table II); these jointly fix the transition matrix. D. SDN Control Plane and Routing The SDN controller is co-located with the carrier and maintains a global view of node geometry, per-link Gilbert–Elliott state, the AMC-selected rate on each 5G link, and the SATCOM availability. The routing policy is per-packet and class-aware. The four elements below describe primary-path selection, failover, encryption of tactical traffic on SATCOM, and the routing modes used for paired comparison. 1) Primary path. Tactical packets prefer 5G if (a) the destination is within band range, (b) the link is in the good state, and (c) SINR supports at least the lowest MCS. Administrative packets prefer SATCOM unless the rain-fade margin is exceeded. 2) Failover. On primary-path unavailability, the controller routes to the alternate path, incurring a configurable controller processing delay Tc .
3) Encryption. AES-256-GCM is applied to tactical traffic crossing SATCOM; the latency contribution Te is recorded separately for the decomposition analysis. 4) Routing Modes. Three modes share the same simulator for paired comparison: sdn (dual-path with failover), static_5g, and static_satcom. E. Performance Metrics Each independent trial reports: Reliability, the fraction of packets delivered; Tactical deadline-miss rate (Dtac ), the fraction of tactical packets exceeding; SATCOM offload, the fraction of delivered packets carried on the satellite path; Latency jitter, defined as p99−p50 of end-toend latency on the 5G path; and Mean latency. For one detail-logged independent trial, latency is decomposed into five contributions τ = τprop + τtx + τretry + τctrl + τenc . F. Experimental and Statistical Methodology The simulation follows discrete-event simulation practice from Law and Kelton [16], using CRN for paired comparison, convergence-validated sample sizes, and parameter sweeps for sensitivity and multi-region analysis. 1) CRN: For each independent-trial index r ∈ {1, . . . , R}, all modes draw fading, rain, Gilbert–Elliott transitions, and traffic from a seeded stream identified by r. The three modes therefore see paired environments, which sharpens SDN-vs-static contrast and is required for paired t-test described below. (base) (sdn) denote a metric and Xr 2) Paired t-test: Let Xr on independent trial r under SDN and a static baseline. We test H0 : E[X (sdn) −X (base) ] = 0 via one-sample t statistic on paired differences Dr , reporting the mean difference, √ 95% confidence interval (CI) D̄ ± t0.975,R−1 sD / R, and t statistic. 3) Sample sizes and convergence: Each parameter setting uses R = 60 independent trials of Np = 50,000 packets, with both choices justified by convergence sweeps on the Reliability metric. Sub-sampling R ∈ {30, 40, 50, 60} from the R = 60 experiment yields 95% CI half-widths on Reliability that drop below 0.1 pp at R = 60. A separate sweep with Np ∈ {20,000, 30,000, 40,000, 50,000} confirms that the mean reliability estimate moves by less than 0.1 pp across this range, well within the per-run Monte Carlo CI. 4) Sensitivity and joint sweeps: Univariate sweeps over Rician K-factor (covering the range observed in over-water maritime channel measurements [22]), ship spacing, and tactical traffic share πtac are run with the same R = 60 independent-trial budget and reported with 95% CI bands. A two-dimensional sweep over the Gilbert–Elliott badstate probability and the Rician K-factor characterizes the joint operating region in which the SDN dual-path advantage is largest. 5) Multi-region study: The baseline configuration is rerun under each of the four regions defined in Section III-A, with the per-region weather record as the only varying input.
TABLE III Per-mode reliability and latency (Norfolk baseline). SatOff (SATCOM offload) shown only for SDN (trivially 0%/100% for static modes). TDMiss (%) 73.97 73.93 100
SatOff (%) 19.77 — —
Mean Latency (ms) 515.29 124.07 612.37
static_5g static_satcom
70 60
∆ (pp)
95% CI (pp)
t
56.87 3.22
[56.79, 56.95] [3.12, 3.31]
1416.88 68.52
75 SDN (dual-path) Static SATCOM
70
SDN
Static 5G Routing mode
Static SATCOM
A. Routing-Mode Comparison SDN achieves 93.09% mean reliability versus 36.22% 5G-only and 89.87% SATCOM-only (Table III). Paired improvements are +56.87 pp over static_5g (95% CI [56.79, 56.95]) and +3.22 pp over static_satcom (95% CI [3.12, 3.31], Table IV). The modest +3.22 pp gain over SATCOM-only understates the operational picture: SATCOM-only misses every tactical deadline (TDMiss = 100%), so its delivery rate is irrelevant for tactical traffic. But SDN’s failover preserves a tactical-capable path SATCOM alone cannot. The marginal increase in tactical deadline-miss rate (sdn 73.97% vs. static_5g 73.93%) reflects this trade explicitly: SDN failover routes 19.8% of tactical traffic to SATCOM, where the GEO transit time alone (≈480 ms) exceeds the 100 ms deadline. The framework thus trades a sub-percentage-point increase in deadline-miss for a 56.87 pp gain in delivery reliability, a worthwhile trade when packet loss is operationally costlier than latency, and one that motivates lower-orbit satellite integration to recover the latency margin. Fig. 2a shows SDN’s per-run reliability above both static baselines across all 60 runs, with no overlap between modes. Mean latency (Table III) for 5G-only is inverse: the lowest mean latency (124 ms) is static_5g because failures do not affect latency but delivery rate. static_satcom has 612 ms mean latency with and no packet meeting the 100 ms tactical deadline. SDN is between the two (515 ms mean) and, as previously noted, offloads 19.8% of delivered traffic to SATCOM, keeping reliability above 93%. SATCOM latency is dominated by the GEO propagation term; the SDN-controller processing delay Tc and AES256-GCM overhead Te together contribute sub-millisecond latency per packet, negligible relative to the GEO transit time on the SATCOM path.
0
5 10 15 Link fault probability πB (%)
20
(a) Per-trial reliability by routing (b) Reliability vs. Gilbert– mode (R = 60). Elliott πB fault probability with 95% CI bands. Fig. 2. Reliability 140 120
80
5G throughput (Mbps)
Tactical deadline miss (%)
90
The configurations detailed in Section III led to the following results. The per-mode metrics are summarized in Table III; the paired SDN-vs-static comparisons are in Table IV.
80
40
100
IV. Results
85
50
30
TABLE IV Paired SDN-vs-static comparison on Reliability (R = 60, percentage points). Baseline
90
80
Reliability (%)
sdn static_5g static_satcom
Reliability (%) 93.09 36.22 89.87
95
90
Reliability (%)
Mode
100
70 60 50 40 30
100 80 60 40
20 20
10 0
0 AMC on
AMC off
AMC on
AMC off
(a) Tactical deadline-miss rate. (b) 5G Throughput. Fig. 3. AMC trades peak throughput for robustness as SINR falls.
B. Behavior Under Varying Link Faults Sweeping the Gilbert–Elliott bad-state probability πB from light to severe degradation, Fig. 2b shows SDN reliability remaining above the static_satcom baseline across the swept range, with 95% CI bands that do not overlap the operating range. The static modes provide visual lower and upper reliability envelopes without path agility. The SDN controller’s failover activates progressively as πB rises, with SATCOM’s share of delivered traffic growing monotonically across the swept range. Total delivery loss is correspondingly mitigated relative to the static_5g baseline. 1) AMC Impact on 5G Performance: The AMC ablation isolates the SDR contribution. The operational benefit is in Fig. 3a: with AMC disabled, the fixed high-MCS scheme misses the tactical deadline on every packet (TDMiss = 100%), so the 5G path is unviable for tactical traffic. Enabling AMC drops the radio to a lower MCS rather than dropping the packet when SINR falls, cutting TDMiss to 74%. The throughput trade is in Fig. 3b: AMC produces a tightly-clustered ≈ 49 Mbps mean rate, while the fixed-MCS configuration shows a higher mean (≈ 112 Mbps) but only on its successful subset of packets and with substantially larger variance. The trade-off is clear: AMC sacrifices peak per-packet rate for predictability and tactical viability. 2) Formation Geometry and 5G Path Utilization: CSG formations are not fixed: screen distances vary with mission phase, sea state, and threat posture. Fig. 4a
TABLE V Per-region SDN metrics; weather sampled from a year of 15-min observations in each region. Location
Reliability (%)
TacDeadlineMiss (%)
SatOff (%)
Jitter5G (ms)
MeanLatency (ms)
PeakRain (mm/hr)
93.09 89.11 91.29 88.42
73.97 74.22 74.40 74.29
19.77 20.68 19.37 20.77
375.77 375.66 375.76 375.64
515.29 506.59 514.42 504.96
77.70 91.80 58.20 87.00
Norfolk NorwegianSea PhilippineSea NorthPacific
35
100
93.2
20
10
0
5
10
Rician K-factor
15
5 0.5
1.5
2
2.5
3
Ship spacing factor × ( baseline)
(a) Rician K-factor sensitivity. (b) Ship-spacing sensitivity. Fig. 4. SDN sensitivity to fading and formation geometry.
shows SDN reliability is robust to Rician K-factor, varying under 0.5pp across K ∈ [0, 16] with overlapping confidence bands; the key result is not tied to a specific channelquality assumption. Fig. 4b sweeps inter-ship spacing multiplicatively over baseline formation: 5G path utilization declines monotonically from ≈ 35% at 0.5× to under 10% at 3×. As ships disperse, fewer pairs remain within high- and mid-band 5G range envelopes, and the controller correctly migrates traffic to SATCOM. 3) Rain-Fade Effects on SATCOM: The per-packet rain-rate sampling and ITU-R P.618 attenuation model [9] produce bit error rate (BER) fluctuations consistent with the predicted slant-path attenuation curve, with BER rising sharply above the rain-fade margin (Fig. 5a). The corresponding hour-of-day reliability dip tracks the empirical rain-rate distribution at each region; reliability is otherwise stable across the diurnal cycle. 4) Joint Operating Region: To characterize where the SDN dual-path advantage is largest, Fig. 5b plots the reliability gain of SDN over the better static baseline as a function of both πB and Rician K. The advantage is largest in the joint stress region (low K, high πB ) and small but positive in clear-sky conditions, which is the operationally relevant pattern: SDN matters most where conditions are worst on both axes. 5) Multi-Region Validation: Holding the CSG topology fixed and varying only the per-region weather record (Table V), SDN reliability ranges from 88.4% (North Pacific storm track) to 93.1% (Norfolk), with the Norwegian Sea polar-edge region at 89.1% and the Philippine Sea at 91.3%. The differences track the per-region peak rain rate more closely than they track latitude. The framework holds across all four regions, and SDN remains the highestreliability mode in each operational region, CI widths remaining comparable.
10-12
3.5
2
10-10
1
4
6.4682
92.5
5.7258
10
4.5
4
92.7 92.6
5
4.9834
10-8
5.5
8 4.241
15
-6
2.7561
92.8
10-4
Rician K-factor
92.9
6
25
BER (log)
93
6.5
10-2
3.4985
5G path utilization (%)
93.1
Reliability (%)
7
16
30
SATCOM 5G
0
10
20 30 40 Rain rate (mm/hr)
50
60
SDN gain vs SATCOM (pp)
93.3
3
0
2.5
0
2
4
(a)
8 12 Fault rate πB (%)
20
(b)
Fig. 5. Physical-layer validation and system-level characterization: (a) BER vs. rain rate on the Ku-band SATCOM link with the ITUR P.618/P.838 prediction overlaid; (b) SDN reliability gain across the joint (πB , K) parameter space with overlaid gain contours.
V. Conclusion We presented a stochastic dual-path SDN/SDR simulator for CSG communications, coupling Rician 5G with AMC and Ku-band SATCOM with ITU-R P.618 rain fade under a Gilbert–Elliott availability model. Paired comparison against static_5g and static_satcom baselines shows that SDN dual-path routing yields significantly higher reliability while sustaining the 100 ms tactical deadline that static_satcom cannot meet, with the AMC element contributing quantifiable robustness as SINR falls. The framework holds across four maritime regions, including a high-latitude polar-edge region. Future work includes adversarial jamming and contested-spectrum scenarios, snow attenuation at high latitudes, lower-orbit satellite integration to recover tactical-latency margins on the failover path, high-fidelity SATCOM geometry computed from GEO sub-satellite longitudes with elevation-angledependent rain fade (capturing the elevation degradation at regions such as the Norwegian Sea and North Pacific), time-correlated rain-rate sampling to model storm persistence, 5G PC5 sidelink for direct ship-to-ship tactical paths, and waveform-level SDR modeling in Simulink. References [1] N. Niknami, A. Srinivasan, K. St. Germain, and J. Wu, “Maritime communications—Current state and the future potential with SDN and SDR,” Network, vol. 3, no. 4, pp. 563–584, 2023, doi: 10.3390/network3040025. [2] C. B. Landis, “A new era of afloat IP services,” CHIPS, October 2016. [Online]. Available: https://www.doncio.navy. mil/CHIPS/ArticleDetails.aspx?ID=8349 [3] R. Akeela and B. Dezfouli, “Software-defined radios: Architecture, state-of-the-art, and challenges,” Computer Communications, vol. 128, pp. 106–125, 2018, doi: 10.1016/j.comcom.2018.07.012.
[4] D. Kafetzis, S. Vassilaras, G. Vardoulias, and I. Koutsopoulos, “Software-defined networking meets software-defined radio in mobile ad hoc networks: State of the art and future directions,” IEEE Access, vol. 10, pp. 9989–10 014, 2022, doi: 10.1109/ACCESS.2022.3144072. [5] D. Kreutz, F. M. V. Ramos, P. E. Veríssimo, C. E. Rothenberg, S. Azodolmolky, and S. Uhlig, “Software-defined networking: A comprehensive survey,” Proceedings of the IEEE, vol. 103, no. 1, pp. 14–76, 2015, doi: 10.1109/JPROC.2014.2371999. [6] M. Cicioğlu, S. Cicioğlu, and A. Çalhan, “SDN-enabled cognitive radio network architecture,” IET Communications, vol. 14, no. 18, pp. 3153–3160, 2020, doi: 10.1049/iet-com.2019.1346. [7] V. Shkarupylo, S. Skrupsky, A. Oliinyk, and T. Kolpakova, “Development of stratified approach to software defined networks simulation,” Eastern-European Journal of Enterprise Technologies, vol. 5, no. 9 (89), pp. 67–73, October 2017, doi: 10.15587/1729-4061.2017.110142. [8] T. Häckel, P. Meyer, F. Korf, and T. C. Schmidt, “SDN4CoRE: A simulation model for software-defined networking for communication over real-time ethernet,” in Proceedings of 6th International OM-NeT++ Community Summit, M. Zongo, A. Virdis, V. Vesely, Z. Vatandas, A. Udugama, K. Kuladinithi, M. Kirsche, and A. Förster, Eds., vol. 66, 2019, pp. 24–31, doi: 10.29007/w71t. [9] ITU, “Propagation data and prediction methods required for the design of earth-space telecommunication systems (ITU-R P.618-14),” Tech. Rep., 2023. [Online]. Available: https://www.itu.int/dms_pubrec/itu-r/rec/p/R-RECP.618-14-202308-I!!PDF-E.pdf [10] Y. Gong, W. Huang, W. Wang, and Y. Lei, “A survey on software defined networking and its applications,” Frontiers of Computer Science, vol. 9, no. 6, pp. 827–845, 2015, doi: 10.1007/s11704-015-3448-z. [11] D. M. Molla, H. Badis, L. George, and M. Berbineau, “Software defined radio platforms for wireless technologies,” IEEE Access, vol. 10, pp. 26 203–26 229, 2022, doi: 10.1109/ACCESS.2022.3154364.
[12] S. Nazari, P. Du, M. Gerla, C. Hoffmann, J. H. Kim, and A. Capone, “Software defined naval network for satellite communications (SDN-SAT),” in IEEE MILCOM 2016. IEEE, 2016, pp. 360–366, doi: 10.1109/MILCOM.2016.7795353. [13] P. Du, F. Pang, T. Braun, M. Gerla, C. Hoffmann, and J. H. Kim, “Traffic optimization in software defined naval network for satellite communications,” in IEEE MILCOM 2017, 2017, pp. 459–464, doi: 10.1109/MILCOM.2017.8170766. [14] Q. Zhao, P. Du, M. Gerla, A. J. Brown, and J. H. Kim, “Software defined multi-path tcp solution for mobile wireless tactical networks,” in IEEE MILCOM 2018, 2018, pp. 1–9, doi: 10.1109/MILCOM.2018.8599848. [15] H. Ghafoor and I. Koo, “Cognitive routing in softwaredefined maritime networks,” Wireless Communications and Mobile Computing, vol. 2020, no. 1, p. 8820838, 2020, doi: 10.1155/2020/8820838. [16] A. M. Law and W. D. Kelton, Simulation and analysis, 2nd ed. New York: Mcgraw-Hill, 1991. [17] E. N. Gilbert, “Capacity of a burst-noise channel,” Bell system technical journal, vol. 39, no. 5, pp. 1253–1265, 1960, doi: 10.1002/j.1538-7305.1960.tb03959.x. [18] E. O. Elliott, “Estimates of error rates for codes on burst-noise channels,” The Bell System Technical Journal, vol. 42, no. 5, pp. 1977–1997, 1963, doi: 10.1002/j.1538-7305.1963.tb00955.x. [19] D. Tse and P. Viswanath, Fundamentals of Wireless Communication. Cambridge: Cambridge University Press, 2005. [20] A. J. Goldsmith and S.-G. Chua, “Variable-rate variablepower MQAM for fading channels,” IEEE Transactions on Communications, vol. 45, no. 10, pp. 1218–1230, 1997, doi: 10.1109/26.634685. [21] ITU, “Specific attenuation model for rain for use in prediction methods (ITU-R P.838-3),” Tech. Rep., 2005. [Online]. Available: https://www.itu.int/dms_pubrec/itu-r/ rec/p/R-REC-P.838-3-200503-I!!PDF-E.pdf [22] J. Wang, H. Zhou, Y. Li, Q. Sun, Y. Wu, S. Jin, T. Q. Quek, and C. Xu, “Wireless channel models for maritime communications,” IEEE access, vol. 6, pp. 68 070–68 088, 2018, doi: 10.1109/ACCESS.2018.2879902.