The design of future wireless networks, particularly in rural and remote areas, faces challenges due to a forecasted surge in connected devices for 6G and beyond use cases, which will strain conventional terrestrial networks. Nonterrestrial networks (NTNs) offer a promising solution, but physical constraints in nonterrestrial environments necessitate multilayered heterogeneous NTNs, integrating layers at different altitudes to balance global connectivity with quality-of-service requirements. Existing studies of NTN performance lack consideration for practical limitations on transmitter availability as well as heterogeneous layer interactions. This work applies stochastic geometry to examine intranetwork interference in the downlink of a heterogeneous NTN-comprised of unmanned aerial vehicles, high-altitude platforms, and low-Earth orbit satellites-while accounting for communication interruptions. The derived signal-to-interference-plus-noise ratio distribution is used to assess terrestrial user coverage probability and the corresponding expected packet delivery delay under a specified retransmission scheme. Numerical results indicate that although increasing layer density initially enhances network coverage, interference ultimately degrades coverage probability. Similarly, while increasing the density of low-altitude layers initially reduces expected delay, further interference eventually leads to longer delays.

Improvements in hardware allow coherent fast frequency hopping (CFFH), enabling tight synchronization and reducing combining losses between hops. To achieve coherence, the timing and frequency offsets must be estimated, but existing solutions are either too complex or perform poorly in low SNR. We propose a noise-robust, low-complexity time-frequency estimation method based on rank-1 lattice sampling for CFFH. The hop sequence design and processing results in a fraction of complexity over existing methods. The tradeoff between complexity and performance is discussed.

Terrestrial networks face limitations, such as restricted rural broadband coverage and service outages during disasters. To address these challenges, nonterrestrial networks (NTNs) are a promising alternative, utilizing aerial vehicles and satellites to enhance coverage and support diverse user applications. However, existing works lack comprehensive consideration of transmitter availability, resource allocation, and cost constraints, leading to suboptimal network performance and design. This work explores the potential of heterogeneous NTNs, including unmanned aerial vehicles, high-altitude platforms, and satellites. Using stochastic geometry, we analyze downlink performance while considering interruptions in communication service for both low-altitude transmitters (due to recharging needs) and high-altitude transmitters (due to fluctuating solar energy harvesting). Our analysis derives connection probabilities, representing the likelihood of transmitters establishing downlink connections with ground users. We propose a resource allocation framework using convex optimization techniques to maximize the downlink connection probability while considering economic costs and communication quality-of-service requirements. Numerical evaluations highlight the significance of incorporating low-Earth orbit satellites and demonstrate the influence of economic cost and signal-to-interference-plus-noise ratio (SINR)-related constraints. Results indicate that allocating resources to higher-altitude layers is favorable under stringent cost constraints while lower-altitude layers are preferred under strict SINR constraints due to improved propagation conditions.

An uplink non-orthogonal multiple-access communication system is developed for heterogeneous non-terrestrial networks (NTNs) that employ both low-Earth orbit (LEO) satellite constellations and high-altitude platforms (HAPs) as relays with hybrid automatic repeat request capabilities. The system is designed as a random-access and interference-resistant network to support low-rate users. A common waveform and frequency band is used for all uplink transmissions (whether the transmitter is a user terminal or a HAP). Both theoretical and simulation-based performance analysis results are provided as a function of user transmit power and packet arrival rates for channels with interference. The use of HAP relays is shown to significantly improve throughput and reduce queue size relative to direct transmission to the LEO satellites, especially for the case of low-power users in high interference environments.

Low-earth orbit (LEO) satellites can enable connectivity for simple, low-power user terminals and have been a significant point of research for upcoming 6G applications. These devices, however, often lack high-power amplifiers and high-gain antennas. Coupling these factors with multipath and shadow fading as well as significant path loss, the achievable signal-to-noise ratio (SNR) at the satellite can be very low in these systems. However, in dense LEO constellations, it is not uncommon for multiple satellites to be in view of a user terminal. By employing joint reception over multiple satellites and an intelligent combining scheme, the received SNR can be substantially improved. Furthermore, time diversity due to the motion of the satellites relative to the user terminal can improve the probability of message reception. In this paper, we quantify the potential performance improvements of different combining approaches, including the effects of different crosslink configurations which may limit the capabilities of combining schemes.

In some channels, such as the frequency-hop channel, the transmission may undergo abrupt transitions in phase. This can require the receiver to re-estimate the phase on each hop, or for the system to utilize modulation techniques that lend themselves to noncoherent detection. How well the receiver can estimate the phase depends on the channel signal-to-noise ratio and how long phase coherence can be assumed. Although prior work has shown that using any reference symbols to aid the phase estimation process is suboptimal with respect to capacity, their presence may be useful in practice as they can simplify the receiver processing. In this paper, the effects of per-pulse phase uncertainty are examined for systems using binary modulation. Both the fraction of the transmission that may be devoted to reference symbols without substantially reducing the overall channel capacity and the point at which it is better to forego coherent processing in favor of noncoherent demodulation are examined.

The use of standard binary turbo codes has previously been shown to work well with noncoherent $M$-ary orthogonal modulation, particularly when used with iterative demodulation and decoding at the receiver. For channels with very low signal-to-noise ratios, the code rate may need to be reduced below those provided by typical turbo codes. Rather than designing additional lower rate turbo codes for these cases, repetition can be used. One possibility to incorporate repetition is to repeat individual $M$-ary orthogonal channel symbols, which can be noncoherently combined at the receiver. An alternative approach is to repeat and interleave code bits prior to mapping them to channel symbols. In this paper, this bit repetition approach, combined with a receiver utilizing an iterative demodulation and decoding technique in which the demodulator itself is also iterative, is shown to outperform symbol repetition.

A new method for generating digital noise-like spread spectrum signals is proposed. A standard binary keystream is used to generate a sequence of chips according to a Gaussian-like chip amplitude distribution for spreading sequences. The properties of these spreading signals are investigated as a function of the number of discrete amplitude levels and number of chips per symbol. The similarities between the generated signals and random Gaussian signals are evaluated based on higher-order moments. Implementation considerations, such as peak-to-average power ratio and amplifier backoff are also considered.

A phased array system with polarization agility for the purpose of interference rejection is evaluated. Polarization agility is defined as a system which can transmit or receive any given polarization and switch its polarization in near real time. First, the efficiency that is sacrificed to gain polarization agility with a simple interleaved dual subarray (IDS) system is quantified. Second, the interference rejection which is gained with a random-polarization-hopping technique is analyzed. Third, various approaches for processing the received signal are discussed and compared. The combined analysis provides a full picture of the true costs and benefits of employing a simplified communication system with polarization agility.

Airborne tactical networks (ATNs) have provided protected air-to-air communications for military aircraft for several decades. To support emerging and future warfighter needs, the next generation of systems will require significant improvements to provide higher capacity, longer range, greater flexibility, and increased interoperability. Governed by domain characteristics such as long transmission ranges, low-to-medium data rates, latency constraints, and link protection needs, the air tactical domain poses several unique requirements on link and network design. Developing next-generation ATNs requires an understanding of the airborne tactical domain, including the design constraints and challenges at various layers of the network stack. In this article, we provide an overview of the unique domain characteristics of ATNs and highlight the key design challenges and research areas associated with the physical, link, and network layers.

Robust frequency hop (FH) waveforms perform forward error correction coding across multiple time-frequency slots (referred to as hops). The typical assumption for FH receivers is that only one signal per hop can be recovered by the demodulator. In terms of multiple-access interference (MAI), this typically leads to one of two design choices for FH systems. One option is to mitigate MAI statistically; namely, through the use of nonorthogonal hopping. With nonorthogonal hopping, from time to time multiple users' signals occupy the same hop. Because coding is performed across many hops, this interference does not necessarily preclude correct decoding, and the corresponding performance degradation may be deemed acceptable. The advantage of this approach is that one user's hopping sequence does not depend on the hopping sequence of one or more of the other users, which reduces the required level of coordination among the users. The second option is to completely avoid MAI through the use of orthogonal hopping sequences. Such sequences ensure that at most one user signal is present in each hop, which complements the capabilities of a single-user detector, but additional coordination is required among the users.

Most traditional coded and interleaved frequency-hop systems either avoid multiple-access interference through the use of orthogonal hopping patterns or tolerate the MAI that results from non-orthogonal hopping patterns. While the former method typically has better performance, all the users' hopping patterns must be coordinated to maintain orthogonality. Because the processing gain of frequency-hop systems relies on the secrecy of the hopping pattern, having a potentially large number of users with knowledge of coordinated hopping patterns may be undesirable. Rather, multiple smaller groups that are cryptographically isolated from each other are preferred. In this paper, we discuss alternatives for supporting multiple such cryptographically-isolated user groups in frequency-hopping spread spectrum systems. Tradeoffs among interference rejection, data rate, and number of supportable users are considered for a variety of channel access schemes and hopping pattern designs. In addition, the impact of improved receiver capabilities, such as multiuser detection, to the trade space is investigated.

Per-hop multi-user detection (PH-MUD) is a general class of techniques for frequency-hopping systems that enable multiple interfering hops to be simultaneously demodulated. Such techniques have the potential to increase the number of supportable users and increase per-user throughput. To realize these gains in actual systems, however, channel estimation capabilities sufficient for successful PH-MUD must be developed. In fact, the fundamental feature of such systems, hopping, requires that estimation be performed on a perhop basis, which precludes training amortization. We describe potential estimation techniques for PH-MUD and evaluate the resulting overhead, energy requirements, and performance. Practical estimation techniques with reasonable overhead are seen to provide good PH-MUD performance; thus PH-MUD remains a promising capability for frequency-hopping systems.

Differential phase-shift keying (DPSK) is commonly associated with a loss in performance when compared to coherent demodulation of PSK signals. Because DPSK avoids the problem of phase tracking, the performance loss may be deemed acceptable. When two or more users transmit simultaneously, such as in a frequency-hopping (FH) system with non-orthogonal hopping patterns, single-user differential demodulation cannot guarantee selecting the most probable hard decision for each symbol. In this paper, we consider the performance when collisions occur. For our technique, we first derive the resulting signal constellation from performing differential demodulation. Conditioned on that, and assuming knowledge of phase difference between the users, the optimal multi-user detection (MUD) algorithm is derived and analyzed to bound the performance of non-coherent demodulation. Based on the analysis, a sub-optimal 2-user MUD algorithm that requires no phase information is proposed and shown to attain significant performance improvement over single-user detection (SUD).

If the channel is particularly poor, a radio may be unable to decode any individual packet reception. Packet combining utilizes retransmission diversity to improve the probability that a message can be decoded. In cases where the transmitter is known to the receiver (e.g., communication using a dedicated time slot), the receiver can request that the transmitter resend the message. However, in random access networks, a receiver may hear packets from many transmitters. If a received message is not decodable, the receiver may not know the identity of the transmitter and thus be unable to request a retransmission directly. The lack of a positive acknowledgment may cause the transmitter to eventually repeat the message, but without any explicit control signaling between the two radios, the receiver may be unaware that the new attempt can be combined with one of its previous receptions. In this paper, techniques to identify which received packets should be combined are considered. The packet matching techniques are limited to those utilizing information contained only in the individual receptions and so are suitable for random-access networks.

Simultaneous transmissions from different nodes in a network can result in multiuser interference in which an intended receiver hears the overlapping combination of two or more transmissions. In the case where there are multiple observations with different time offsets of a collision (e. g., if the packet retransmissions also collide, or if there is a cooperating receiver with different propagation delays to the transmitters), zigzag [1] decoding can be employed. In this paper, we analyze the performance of zigzag decoding for the case of two receiving nodes with two simultaneous transmitters in an additive white Gaussian noise (AWGN) channel. We also present a soft-decision version of zigzag decoding and show that it is the optimal MAP decision rule.

A typical military airborne networking scenario involves small groups of relatively closely-spaced aircraft which fly as a unit to perform a common mission. These aircraft need to maintain very long data links with other aircraft in the presence of strong interference from hostile jammers. Due to the physical separation of the aircraft, the time difference of arrival can often be greater than the chip rate of the desired signal, allowing multiple aircraft to perform interference cancellation to greatly increase their resistance to jamming. In this paper, the performance of a zigzag-like algorithm [1] is simulated and compared against single receiver performance in the presence of strong interference. Performance of these algorithms in the presence of strong interference is shown for both uncoded and coded transmissions, which effectively mitigate strong interference with only 0.5 dB of performance loss compared to the unjammed case regardless of the jammer strength.

Frequency hopping provides some protection against multiple access interference in random-access ad hoc networks. Power control and the use of short routing hops can further reduce interference by promoting spatial reuse. If the network is fully connected, frequency hopping and power control alone may be sufficient to allow good throughput without the use of routing. This approach is particularly desirable as avoiding routing can simplify network setup and reduce packet delay. However, if the network is heavily loaded, frequency hopping may no longer sufficiently protect against interference, so including routing may be beneficial. In this paper, we consider application of least interference routing (LIR) to frequency hop random-access networks. The performance of a family of LIR metrics is considered for a variety of networks to determine which cases multi-hop routing should be used, and it is shown that the choice of the optimal metric is sensitive to several network characteristics.

The effects of adaptive-rate transmissions and routing on the total throughput of a slow-frequency-hop packet-radio network are considered. Adaptive rates are achieved through the use of error-control coding with perfect side information. Both fixed-length codes and variable-length codes are considered. Performance results are obtained for direct transmission (i.e., no routing), two-hop limited routing, and full routing. Each link metric is a function of the amount of interference caused by using the link. We compare the total network throughput for each of these schemes with both fixed-rate and adaptive-rate coding.

For a frequency-hop (FH) spread-spectrum communication system, the packet error rate is dependent on both the performance of the error-correcting code and the ability of the receiver to detect and acquire timing of incoming packets. It is important that the error rate for one does not limit the overall error rate of the system. Turbo coding has been shown to be robust in the presence of partial-band interference, but if the acquisition scheme is not able to determine the presence of a packet, the coding will not help system performance. Rather than focusing on optimal signal acquisition methods, this paper focuses on computationally simpler packet acquisition relying on hard decision combining of individual pulse detections. The performance of this acquisition strategy in the presence of partial band interference (PBI) is compared to that of several turbo coded systems in both additive white Gaussian noise (AWGN) and Rayleigh fading channels.