Robust and reliable routing decisions are enablers to minimize redundancy and to optimize throughput. This might be nice to have in non-time critical transmissions but required in QoS related data delivery, like VoIP and video calls. MANETs, are error-prone considering their wireless and mobile nature. Relying on an up-to-date and complete topology overview makes routing easier and more efficient and also facilitates mentioned routing goals. MANETs must consider interference if routing follows Ford-Fulkerson's objective to compute maximum throughput. However, preconditions must be met before one optimizes flow distributions. Such major precondition in the field of MANETs is an algorithm that constructs a graph equal to the actual MANET topology. This graph must be complete and almost instant in terms of connections and construction to guarantee that each node's position does not deviate from its actual. We centralize the routing, inspired by SDN advancements, to a controller that obtains the topology by triggering an algorithm, called RFTKR. Nodes assemble a tree, having the controller as their root while gathering neighborhood information. Afterwards, they convey their adjacent lists towards the controller. We present a proof-of-concept evaluation with microcontrollers and discuss the scalability of RFTKR with an extensive simulation.
With an increased adoption of uninhabited platforms in tactical environments such as search and rescue operations the proliferation of tactical assets equipped with advanced computing, sensing and networking capabilities becomes complex to manage. Edge/tactical cloud architectures provide the necessary automation processes for managing and scaling services and applications deployed across a larger number of platforms. However, size weight and power restrictions on tactical platforms leads to a heterogeneity in deployed infrastructure for both compute capabilities and available network bearers which raises complexities that need to be addressed by the underlying cloud infrastructure. In this paper we propose and demonstrate a heterogeneous cloud architecture used for a search and rescue training operation conducted by Surf Life Saving South Australia in collaboration with the Defence Science and Technology Group.
The main technologies currently used for Space Domain Awareness (SDA) are radar, optical telescope, and reflected laser tracking. These technologies alone cannot provide the comprehensive and persistent SDA that is becoming increasingly vital due to the increase in space traffic and the resulting higher probability of collisions. Passive Radio Frequency (RF) sensing is a promising technology to support the SDA mission, particularly due to its ability to operate 24/7 in all weather conditions. By observing the Doppler shift and angle of arrival of signals transmitted by satellites, their orbit can be determined. This paper describes an algorithm for Doppler estimation and demonstrates its efficacy for Low Earth Orbit (LEO) satellites. The algorithm can effectively estimate Doppler for both continuous and intermittent satellite transmissions.
Emulation environments are an effective approach to experimenting with and evaluating network protocols, algorithms, and components. However, implementing a complete waveform for an emulation environment requires a significant effort. This paper describes two simplified approaches to implement and include a Synchronized Cooperative Broadcast (SCB) waveform within the Anglova scenario and the EMANE network emulation framework. The result is two different implementations of SCB, having different levels of abstraction regarding the specific details of SCB and offering different capabilities. The SCB implementations are tested and compared in the militarily-realistic Anglova scenario. The performance of SCB is also theoretically estimated within the Anglova scenario as a baseline for comparison purposes. The results show that the approaches taken to emulate SCB within EMANE and the Anglova scenario perform as expected when compared to the theoretical results.
To minimise the probability of detection of a military network the power received at a distant point should be reduced. This can be achieved in various ways. Topology control has been shown to be able to reduce transmission power in these networks, lowering the network's RF power footprint and so the probability of detection. This paper looks at how positioning autonomous vehicles can also be used to further reduce the received power at a adversary position and hence the probability of detection. An algorithm for this positioning is described and results presented.
Experimentation focused on assessing the value of complex visualisation approaches when compared with alternative methods for data analysis is challenging. The interaction between participant prior knowledge and experience, a diverse range of experimental or real-world data sets and a dynamic interaction with the display system presents challenges when seeking timely, affordable and statistically relevant experimentation results. This paper outlines a hybrid approach proposed for experimentation with complex interactive data analysis tools, specifically for computer network traffic analysis. The approach involves a structured survey completed after free engagement with the software platform by expert participants. The survey captures objective and subjective data points relating to the experience with the goal of making an assessment of software performance which is supported by statistically significant experimental results. This work is particularly applicable to field of network analysis for cyber security and also military cyber operations and intelligence data analysis.
Combining several mobile networks in order to build a federated network at the tactical edge is a challenge because of the high degree of mobility and limited data capacity available in these networks. Providing the different partners in a coalition with direct connectivity at the tactical edge can be beneficial since it allows for better cooperation between capabilities from multiple nations and shortens reaction times compared to traditional hierarchical communication models. This paper describes an experiment performed at the Coalition Warrior Interoperability eXercise (CWIX) 2019 with a Depth First Search (DFS) routing protocol that can be used as an inter-network routing protocol to build a federated network. Norway, the Netherlands and Germany participated in proof of concept tests of the routing protocol that showed that the protocol was functionally able to provide connectivity in all the selected use-cases. The results of the tests will be input to discussions regarding future spirals for the Federated Mission Networking (FMN) specifications.
High Frequency (HF) communications is widely utilised in military, safety and amateur radio systems. Utilising HF communication channels for TCP/IP is highly constrained due to the characteristics of the channel which are presented as low link reliability, limited bandwidth and high transmission delays. Low link reliability is the effect of error prone behavior of HF channels which are governed by physics and environmental conditions. Various implementations of Wideband HF (WBHF) provide additional bandwidth but the low reliability still pose a challenge for TCP/IP communications. In this paper, we present a novel TCP named HF-TCP which is an enhanced version of the standard TCP to provide better performance on HF communication channels. HF-TCP modifies key parameters of TCP to allow TCP sessions to remain open over an extended period of time while the low-data is transmitted with induced delays. Results from experiments show improvements in transmission success rate of 24 percent over standard TCP which failed to initiate or maintain communication. The proposed HF-TCP is backward compatible with generic Local Area Networks (LAN) and wide Area Networks (WAN).
It is expected that future tactical warfighting functions, such as tactical picture management and sensor tasking, will be distributed applications running on different systems, connected via a network. Recently, an air force fifth generation network has been proposed where airborne networks are linked with other domains to form a multi-domain network supporting a combat cloud. There are many challenges in building such networks. We propose in this paper a software-defined multi-domain network (SD-DMN) architecture that provides resilient multi-domain networks. Our proposed architecture consists of a software defined communication system to integrate multiple bearers and a Border Gateway Protocol (BGP) control plane for connectivity, traffic and policy management. Using this unique architecture, we develop an algorithm provides multiple reliable transmission paths between any two nodes. Network emulation experiments have been conducted using the Mininet emulation software. The Mininet experiments and prototype implementation validate the feasibility of our proposals.
For most military land vehicles, communication systems are the main contributor to RF emissions from the vehicle. To reduce detectability of the vehicle, the transmission power of the communications systems should be managed. Accomplishing this without adversely effecting the ability of vehicles to communicate requires a coordinated approach. Previous work has developed algorithms that aim to use distributed connectivity control to minimise the received power of network communications at an estimated adversary location. This paper builds upon that work. It proposes a number of possible enhancements that aim to reduce the optimisation time of distributed connectivity control algorithms. The proposed enhancements include partitioning the graph using articulation points to reduce the search space, the ability to eliminate multiple links at once and the elimination of redundant bids in the auctioning process. This paper shows in simulation that these enhancements, combined, significantly reduce the optimisation time compared to the baseline approach.
In this paper we examine computational workload modelling in a generic maritime combat system. We show how to construct models so that executable modelling can then be used to experiment with different hardware approaches for cost, power or performance improvements. This can assist in identifying problems earlier in the design lifecycle than by using traditional design methodologies such as prototyping. We use a generic sonar suite as an exemplar, showing the considerations required in building executable models for traditional and adaptive beamforming algorithms.
Software-Defined Networking (SDN) is a relatively new technology that enables network operators to have very tight but also very flexible, policy-defined control over traffic flows in the network. Using SDN in Time-Sensitive Networking (TSN) offers network operators not only control of the network but also manageability of key network parameters such as latency, throughput, and reliability. In this paper, we discuss the state-of-the-art research related to SDN-enabled TSN networks, present system requirements for time-sensitive heterogeneous wireless distributed software-defined networks, and derive a system architecture. The proposed system architecture not only enables large distributed heterogeneous TSN networks, but also integration with other (legacy) systems, and network management by multiple operators.
Space., Weight and Power-Cooling (SWaP-C) are major design concerns for resource constrained platforms. With modern combat system designs increasing in complexity., approaches that provide early insight into design choice impacts becomes important and critical for design and risk mitigation efforts. In response to this and an identified modelling and analysis capability gap for combat system integration and performance evaluation., DST Group has developed a new type of measurement-based modelling and analysis environment. Built on an approach of constructing the actual system computing infrastructure and deploying models representing application behaviours., measurement-based analysis can provide for early insight in integration and performance risks associated with combat system design choices. Furthermore, this approach provides insight earlier and at a higher fidelity than traditional modelling approaches.
We detail the use of open source training environments to investigate the applicability of standard reinforcement learning techniques to inherently error prone tasks expected in real world application of artificial intelligence. Numerical experiments were conducted in which the performance of both Q Learning and Policy Gradient agents' ability to obtain high reward was compared as the observation state measurement uncertainty was increased. The purpose of the research was to assess the applicability of reinforcement learning to real world applications of self-protection of military platforms, where it is expected that the observed state space is uncertain at best. We found in our experiments that Q Learning is more stable in the presence of state uncertainty than policy gradient learning.
Complex military systems are typically cyber-physical systems which are the targets of high level threat actors, and must be able to operate within a highly contested cyber environment. There is an emerging need to provide a strong level of assurance against these threat actors, but the process by which this assurance can be tested and evaluated is not so clear. This paper outlines an initial framework developed through research for evaluating the cyber-worthiness of complex mission critical systems using threat models developed in SysML. The framework provides a visual model of the process by which a threat actor could attack the system. It builds on existing concepts from system safety engineering and expands on how to present the risks and mitigations in an understandable manner.
Model Based Systems Engineering (MBSE) is accepted as a key enabler for evaluating requirements and designs of combat systems. Systems Execution Modelling (SEM) is an MBSE approach that allows the software system to be modelled independently from the target hardware, inferring hardware characteristics by direct stimulation and measurement. SEM builds a system model out of simple workers, and this paper proposes enhancements to existing workload models to better support the evaluation of combat systems in heterogeneous compute environments comprised of CPUs and GPUs.
Local Area Network (LAN) workstations that operate at the edge tier of Industrial Internet of Things systems (IIoT) and have direct or indirect interaction with critical control devices could be a key vector for advanced threats against control systems, such as a ransomware threat. This indicates that there is a necessity for monitoring these workstations to detect any malicious behavior related to ransomware, and generating an alarm to prevent the ransomware from expanding its activity to more critical system entities. The efficient detection of a ransomware attack very much relies on how accurately its activities are understood and how its traits are discovered. This can help in distinguishing ransomware from legitimate system activities. In this paper, we utilize deep learning techniques to extract the latent representation of a high dimension of collected data to identify malicious behavior accurately. Specifically, the model that we propose is based on a hybrid feature engineering technique of classical and variational auto-encoders. This hybrid technique is used to reduce the dimension of data and extract a good representation of the collected system activities. Then, the new feature vector is passed to a classifier that is built based on deep neural network and batch normalization techniques. The paper concludes with experimental results demonstrating that our model performs better in detecting ransomware compared with other existing models.
Most satellite communications monitoring tools use simple thresholding of univariate measurements to alert the operator to unusual events [1], [2]. This approach suffers from frequent false alarms, and is moreover unable to detect sequence or multivariate anomalies [3]. Here we consider the problem of detecting outliers in high-dimensional time-series data, such as transponder frequency spectra. Long Short Term Memory (LSTM) networks are able to form sophisticated representations of such multivariate temporal data, and can be used to predict future sequences when presented with sufficient context. We report here on the utility of LSTM prediction error as a defacto measure for detecting outliers. We show that this approach significantly improves on simple threshold models, as well as on moving average and static predictors. The latter simply assume the next trace will be equal to the previous trace. The advantages of using an LSTM network for anomaly detection are twofold. Firstly, the training data do not need to be labelled. This alleviates the need to provide the model with specific examples of anomalies. Secondly, the trained model is able to detect previously unseen anomalies. Such anomalies have a degree of unpredictability that makes them stand out. LSTM networks are further able to potentially detect more nuanced sequence and multivariate anomalies. These occur when all values are within normal tolerances, but the sequence or combinations of values are themselves unusual. The technique we describe could be used in practice for alerting satellite network operators to unusual conditions requiring their attention.
Land tactical radio networks are used in military operations to create communication channels for tactical units at the force's edge. These radios must work in a dynamic, contested environment where bandwidth is usually at a premium due to the limited radio capabilities and the many demands placed on the system. It is important therefore to manage the networks as efficiently as possible. Measuring the state and performance of the network is required to achieve this, which is a non-trivial exercise due to the complexity and variability of network designs. Any measurement schema must take into consideration the implementation, state and use of the radio network. Previous work in this area has concentrated on commercially available networks rather than the bespoke networks that are often employed in the military. As part of the SMARTNet research programme, this paper looks at techniques that are currently available for measuring a tactical network's performance, investigates what other work needs to be done, and suggests a way forward. A simple network has been analysed mathematically and then emulated using the Enhanced Mobile Ad-hoc Network Emulator (EMANE) to highlight the issues discussed in this paper.
This paper presents a novel passive tracking system to localize a moving target using asynchronous self-locating receivers, which passively listen to the IEEE 802.11 signals transmitted by the target. We have developed passive single-antenna IEEE 802.11ac receivers that estimate time-difference-of-arrival (TDoA) between multipath components at each receiver, and as such, they do not depend on synchronization between the target and the receivers. We have developed a new localization algorithm based on particle filtering (PF), which completes its execution in real-time within 1 s for each target location to be estimated. The performance is demonstrated experimentally and shown to have a target localization error below 30 cm for all the target locations.
