Dependability in Future Battle Network System — Transport Layer Ability to Maintain Quality of Service

Modernization of armies is a constant process and is driven by intuitive fact that those who do not modernize will become extinct. In last five decades, the development of modern armies has taken place around Colonel John Boyd’s theory of OODA loop that deals with information superiority. Building a robust, mobile and capable network that could provide for novel appliances and information superiority is the main challenge which modernizers are facing. Network, suitable for future combat operations, and able to transport a vast amount of information on a battlefield, is expensive to build. Every mistake in design and the need to correct those mistakes could halt development in an army for years. Therefore, system dependability analysis during system design phase is needed. In this report, the concept of a future Battle Network System is described. The Report evaluates operational environment of BNS and possible failure reasons of the service, and illustrates the change in BNS Quality of Service due to probable transport layer errors. This paper describes the method of testing the concept of proposed network systems on the drawing board, and emphasizes design points for a new system. Nevertheless, the proposed method is by no means conclusive. Rather, it describes an engineering approach to define the main problems while creating MANET-based networking systems.


Introduction
The concept of future Battle Network System has been proposed. Battle Network System (BNS), also known as Combat Systems Network [1] is described by five How to cite this paper: Dieves, V. (2016) Dependability in Future Battle Network System-Transport Layer Ability to Maintain layers: sensors and platforms, services, applications, transport layer and standards. Transport layer itself is described as a system consisting of eight subsystems: 1) Tactical airborne subsystem; 2) Maneuvre leader subsystem; 3) Combat radio subsystem; 4) Cable network subsystem; 5) Mounted local subsystem; 6) Data distribution subsystem; 7) Tactical trunk subsystem; 8) Command-post module subsystem.
In addition to that system, Quality of Service (QoS) benchmarks have been described. In order to evaluate if transport layer is able to reach those benchmarks, one thing has to be understood that the transport layer must maintain Every subunit is equipped with one or two nodes. That means the whole Brigade has approximately 500 nodes.
Above the Brigade, there is one more level of units that have to be connected to BNS (operational level). Every subunit's level has its own set of services which it has access to. Quality of Service benchmarks have been described for every service point in three categories: time tolerance of transmission, tolerance for mistakes and data transmission speed. Table 1 describes QoS benchmarks in BNS.
The concept of Battle Network System (BNS) determines network on that area is mainly created as Mobile Ad-Hoc Network (MANET). Main subunits of BNS are maneuver leader subsystem, combat radio subsystem and tactical trunk subsystem, all of them working as a MANET. Therefore, previous analysis of MANET could be used to evaluate BNS. Tactical trunk subsystem acts as a backbone of BNS. It consists of nodes capable to form connections with each other similar node over the AO of Brigade. The concept states that nodes forming tactical trunk subsystem have to be operational within 15 min after arriving to a new position. All other subsystems connect through a gateway to tactical trunk subsystem. Combat radio subsystem is the main subsystem used by maneuver units. It consists of nodes carried by fighting vehicles, sometimes also by soldiers (signalers). Combat radio subsystem forms local MANETs that are connected with tactical trunk subsystem through a gateway to residing on company level. Maneuver leader subsystem is an extension of combat radio subsystem. It allows a unit commander to dismount and lead his or her subordinates while still receiving all services of BNS.
MANET is a network concept emerged from earlier mesh-network concept. It is a continuously self-configuring network of mobile devices that are connected wirelessly. MANET nodes are able to move randomly in an operational area, while still maintaining a working network. Data packets are exchanged between different nodes using hop. Point-to-point connections between sender and receiver are not needed, though BNS concept states when low delay is needed, e.g. with voice and video subservices, BNS will try to create a point-to-point connection.
QoS in MANET is always "soft", meaning that with constantly changing environment and network architecture, it is not possible to ensure continuous service to all nodes: some of nodes will be at least some time out of service. Therefore, it is not easy to express dependability in MANET with solid number. There are other problems with MANET like path loss, multipath fading, link outage probability, node outage probability, shadowing, hidden terminal problem, exposed terminal problem, etc. Battlefield is a complex environment with many emerging problems, but due to the fact that BNS is described only on concept level, it is not feasible to analyze all aspects of dependability of the system. In this report, node outage problem, hidden terminal problem and those problems impact on QoS will be discussed.
Due to the features of existing military MANET radios, one network can consist of maximum 30 -40 nodes. In order to merge those subnetworks together and work seamlessly, one extra node (gateway) has to be placed on subunits level. Every one of those gateways serves as a critical node for the network. The node is critical if its sudden failure divides a network into two disconnected sub networks [2]. The report will discuss critical links on a battlefield and how this fundamentally affects QoS in BNS.
Throughout recent military history (last 200 years), one certain characteristic has described developments on the battlefield, namely the spreading out of troops and fighting platforms. Spread has several advantages: 1) It is more difficult to determine the position of troops by the enemy; 2) It is more difficult to attack identified troops and cause them excessive damage by the enemy; 3) It is easier to conduct synchronous maneuver with own troops if there is maneuvering space.
The main disadvantage of spread is the weakening of communications. Spread of troops will make it harder to form point-to-point links between nodes allocated to the troops. The same is also true in BNS. The concept describes AOs for different subunits and propagation of signals. As shown in Table 2, on platoon level and up communication is near-line-of-sight or beyond-line-of-sight.
Those parameters should be kept in mind while evaluating QoS in BNS.

Dependability on a Battlefield
Dependability of systems on a battlefield has important implications. Every mistake in system design or operations could have catastrophic consequences if a necessary service is not delivered. Therefore, before analyzing dependability of the BNS, one has to understand what a battlefield is, how it is organized and what are the main threats regarding the system service delivery continuity perspective.
The conventional battlefield is continuous and linear [3]. It consists of securi- Units' concentration on a battlefield varies. Spread is greater in security area and rear area. In main battle area, concentration of forces is greater; therefore, also concentration of nodes is greater. Figure 1 described that spread difference.
Dots are representing clusters of nodes forming local MANETs. As figure shows, the spread of units and nodes lessens in the second quarter (counted from the right side) of AO (represented by rectangular out-shape) and starts then to widen again. Usually MANET architecture describes also the architecture of Internet in order to describe a gateway. That is not so with the concept of BNS. BNS will be built in a way that it is capable of working without Internet connection independently (although BNS can use some services over Internet).
NCC services described in BNS concept are providing situational awareness to the units. That part of BNS will form Battle Management System or BMS. There are several problems on a battlefield that BMS is designed to overcome. In this report, we will discuss target engagement (here in the meaning of processes taking place on tactical level before and during engagement with fire), close maneuver of different subunits and indirect fires coordination.
Target engagement and targeting are closely related. On a tactical level, those include identification, selection and prioritization of targets, combat identification, and engagement with fire and battlefield damage assessment [4]. As Joint Hidden terminal is a well-known problem of MANET. It is essentially an inability to receive a signal from one node by another node because third node's is already sending a signal to the second node, while unaware of another signal being sent [5]. We could easily imagine that on a battlefield, where signal paths vary and randomly change, hidden terminal problem is constantly affecting friendly forces tracking service, and therefore, safety of own troops, especially in time critical situations. Let us picture a hypothetic situation where squad leader tries to identify friendly and enemy vehicles on a battlefield. Squad leader spots an APC (Armored Personnel Carrier) in distance and is not sure if that happens to be friendly or non-friendly weapon platform.
In order to answer that question, squad leader will check his/her BMS interface. If that actually happens to be a friendly unit whose friendly forces tracking subservice has failed to update because of hidden terminal problem, squad leader will identify it as an enemy unit and will engage it with fire. MANET problem has led to catastrophic consequences.
Maneuvering of subunits is another example where BNS dependability is critical. Closely moving and fighting subunits whose maneuvers are not coordinated In nowadays battlefield, forces that lose capability for high tempo will become an objects of targeting process and could be easily engaged with fire, e.g. as it happened with Ukrainian troops in the Battle of Ilovaisk 2014 [6].
Indirect fire coordination is one of the most time-critical subservices of BNS.
It consists of identifying a target, identifying friendly forces near the target, selection of weapon system and firing method, engagement with fire and battle damage assessment (BDA). All well planned and conducted engagements with fire are synchronized with maneuver. Indirect fire is used to suppress enemy forces while maneuver unit closes into striking distance or conducts a maneuver to a better firing position in order to destroy targeted enemy unit. Indirect fire is used to suppress enemy's units until last moments before decisive strike, to minimize its ability to fire back or to take any countering action. That means indirect fire is used in close distance of own units.
In a battle, changes in environment and enemy actions could lead to changes of friendly unit's plans. Therefore, maneuver units have to be able to coordinate the use of indirect fire, e.g. prolong firing on enemy's position if own units movement is halted due to an obstacle, or stop fire if cap between striking position and enemy unit has closed quicker than anticipated, or move fire to previously unexpected and unplanned target. Every change in indirect fire plan has to be communicated to the firing unit without a significant delay. If a maneuvering unit sends out a correction to change fire plan but that correction is not received by the firing unit, the situation could result with a discord of a maneuver plan and a fire plan, leading to excessive losses of combat force. In  MHz. Again, for practical reason we have to take into comparison with other radios the smallest bandwidth value [8].
Streamcaster 4200 (Figure 4) is a radio designed for low intensity conflict environment. It has currently only rudimentary functionality and would need further product development in order to suit for a battlefield. Nevertheless, Streamcaster 4200 promises to guarantee more than 100 Mbps throughput while maximum channel bandwidth is only 20 MHz. Currently smallest bandwidth in  use with Streamcaster 4200 is 5 MHz, although the company producing the radio has already acknowledged that smaller bandwidth is needed and development towards that is taking place [9]. R&S M3TR ( Figure 5) is a military MANET radio family that is considered to be one of the best of its kind on the market. M3TR is software defined multiband radio covering frequencies 1.5 -512 MHz, with programmable bandwidth. Compared with other radios, M3TR data rate is lower-up to 72 kbps. In realistic, BNS that kind of low data rate could be an important restriction [10]. Table 3 shows comparison of different radios average maximum communication distances (calculated from power output or taken from information provided by producing companies) and their responding bandwidth or data rates.
As Table 3 shows, all MANET radios would fit into AO of company (as showed in   1) prioritizes traffic on transport layer for safety-critical subservices; 2) optimizes channel use for different subservices; 3) restricts subservices on the basis of need.
Optimization means that there is probably low resource left in the system for sudden load surge if some nodes are going out of service-result of node outage problem. Therefore, node outage problem in BNS has to be addressed.

Analysis of System Dependability
Node outage problem is a well-known MANET problem. It is defined as a situation that occurs when a node stops its function. In the case where a cluster leader stops functioning, the sensor network protocols should be robust enough to mitigate the effects of node outage by providing an alternate route [12].
There are many ways to calculate MANET reliability, in this report terminal-pair method perfected by M. Ahmad and D. K. Mishra is used in order to describe reliability change in BNS. Terminal-pair reliability is defined as the probability of successful communication between any two selected terminals. BNS could be imagined as a sequence of a number of NLNs, where NLN is nodelink-node connection. Those nodes in NLN have some independent working/operational probability [2]. In NATO countries, unit or set of fighting platforms is considered to be destroyed (defeated) if that unit has suffered more than 30% of losses. Let us assume that if military unit is destroyed, so is node corresponding to that unit. Destroyed unit would be removed from a battle for re-staging. Therefore, if we are considering only reason for node to stop working a damage during a battle, node operational working probability (Pn) is greater than 2 3 .

1 3
Pn Link existence probability affects NLN probability, being dependent on operational status of connecting nodes. Link is operational when two nodes are in each other's transmission range. Therefore, the probability of link existence is the probability that two nodes reside in each other's range [2]. Probability for any possible link (Pl) is between 0 and 1.
Probability of any NLN (Pnln) is:  In this report, it is not feasible to calculate overall reliability for BNS due to the fact that Brigade is a very large unit with lot of subunits (model of Infantry Brigade consists of 278 important agents, excluding most of combat service support subunits [11]). Prior to the calculation of configuration of reliability, configuration model has to be created. For that purpose every subunit and every node has to be described. This is not possible due to the fact that BNS itself does not exist yet. Therefore, in this report set of local MANETs is described for reliability calculations, in order to illustrate QoS change in BNS during an operation.
With a network comprising of ~500 nodes, E (total number of all possible NLN sequences) would be equal to 124,750. But as stated before all those nodes will not be in communication distance of each other. For reliable calculations smaller network, local MANET has to be described. Let us assume that there are two maneuver companies operating side by side. Both are with equal set of nodes. Companies belong to different battalions; therefore, they use normally a different route to the BNS. Let us also assume that both of those companies have critical link to overall BNS through gateway operating on company level ( Figure   6). In

Pc t =
If we set Pl value to 0.75, then: with critical node leads to higher load, which in return leads to higher failure rate. That means that more hops there are in a particular system, less reliable it is.
Hypothesis can be verified by changing "a" value. If  ( ) Pc t change in BNS. require in a system affects reliability of a system.
In our example of two companies, both company's local MANET has E = 45 different ways to configure the network. Let's assume that edge-node has to take minimum two hops to reach a gateway (due to the change in environment). If in this network amount of NLN connections shrinks, then at some point a new critical link will be formed. Pl value will lessen, therefore, lessening configuration reliability.
D. Zhang and J. P. G Sterbenz research [14] shows that with the increasing number of simultaneous node failures, the performance difference between critically linked part of a system and well linked part of a system becomes smaller.
Therefore, one could conclude that as many nodes of BNS as possible should be able to form a new gateway to tactical trunks subsystem or to neighboring clusters.
Concept of BNS implies that node outage happens not only due to the damage, but also when tactical trunk subsystem node changes its position. In this case traffic from corresponding companies has to be diverted to neighboring gateways. That in turn will increase error rate of transport layer, because errors mostly occur when alternative routes exist but are relatively long, and therefore, may not provide satisfactory service in application [15].
As one could see, node outage problem has a major effect on QoS in BNS.
Coming back to safety-critical dependability issues on a battlefield, one has to ask how this affects our understanding of a reliability of a system. As an example with local MANETs and two companies showed, with many critical links in a system overall reliability of a systems ability to provide safety-critical subservices for every node in a system is relatively low. One could say that due to the fault in system design, error that could lead to catastrophic failure is bound to happen. Therefore, it is this report's authors opinion that even in future battlefield Fire Support Coordination Measures have to be in place, separating and de-conflic-ting maneuver and fire of different subunits. Another option would be to calculate every node's possible position, taking into account where its last coordinates were, what is its practical possible speed, etc. But this could mean that a new subservice has to be created in BNS service layer, increasing the burden on transport layer even more.
Increase of critical links increases work load and could disrupt BNS ability to manage traffic in transport layer. One has to keep in mind that due to service layer and transport layer data rate differentiation protocol developed for BNS is probably already using transport layer resources as efficiently as possible. Thus protocol should be able to react to significant change in transport layer. Due to limited resources of transport layer, service differentiation has to be created and safety-critical subservices priority has to be granted.

Conclusions
In this report, the concept of future Battle Network System was introduced. Due to the fact that most of BNS transport layer could be described as MANET, the report evaluated BNS transport layer's ability to maintain Quality of Service through known MANET problems, using MANET evaluation analysis techniques. Although report findings could be counted as intuitive truths, in author's opinion, they still have a value for system developers, namely helping to state the problems and steps to solve those problems.
MANET is probably the best choice for BNS transport layer. Nevertheless, MANET problems are going to affect BNS and have to be addressed during the development phase. First, every MANET is bound by critical links. Therefore, as many nodes as possible should be able to form a gateway to tactical trunk subsystem.
Second, nodes should be with as long communication distance as reasonably possible. That will decrease the amount of hops in BNS, therefore, decreasing error rate. Radios with longer communication distances are also able to form NLN sequences with neighboring local MANETs, thus decrease the number of critical links in the system. Third, safety critical subservices in BNS have to have a priority and have to be superimposed by battle planning rules and Fire Support Coordination Measures.
The hope, that BNS service layer of Network Centric Command (e.g. BMS) would allow to de-conflict fires and maneuvers with high accuracy and increase the efficiency of battlespace use, is not feasible without significant risk-taking.
Fourth, subservices in BNS should be restricted in the basis of need. That would lessen the burden of transport layer and would simplify the development of communication protocol.
Further research is needed to determine, what would be: 1) The suitable protocol for BNS; 2) Reasonable bandwidth in realistic spectrum environment; 3) Possible alternatives to ensure that errors in safety-critical subservices would not lead to a failure.
For the next step, parts of BNS transport layer and service layer have to be created. Transport layer could be a rudimentary set of nodes, which it does not need to represent the whole BNS. But in author's opinion, the service layer should be as complete as possible in order to be able to provide realistic traffic for transport layer. Then a realistic scenario for simulation has to be created.
Scenario should not only provide traffic, but also address different errors. Different sets of node outage problems have to be tested during simulation. Measurements will show possible faults in BNS architecture, allowing making corrections during system design phase.