Integrated Defense Network
Abstract
System of Systems simulation of an integrated defense network. This simulates the interaction between various types of sensor systems (green) and weapon systems (blue), accounting for network architecture and latency, sensor frequency and detection probability, weapon accuracy, lethality and shot flight time. The system models engagements from various quantities of hostile aircraft.
Nodes
Weapon and sensor nodes represent completely functional beta-level systems, i.e., all the hardware, software, personnel, support infrastructure, fuel, ammunition, data and network communication hardware, in short, everything required for the system to perform its operational role.
Network
Network topology is displayed with pink lines. These lines indicate lines of communication between different nodes. When a node is destroyed, communication can no longer route through the node. The network can be grown using uniform or scale-free growth models. Network density can be set from 0 (indicating the minimum connectivity possible to connect all nodes) to 1 (indicating maximum connectivity where all nodes are connected to all other nodes). Network latency can be set from 0.1 and up. This enables the simulation of low-latency networks such as computer networks, or high-latency networks such as voice-over-radio and other human-in-the-loop systems.
Communication
Network traffic is not displayed, but is modeled. Sensors generate "detection" messages, which have a transmission delay. After the delay, the message is routed to the most suitable weapon to engage the threat identified by the sensor. The routing accounts for network latency, as well as the path from the sensor to weapon. The weapon is selected using a heuristic that accounts for factors including weapon reload status, ammunition remaining, and an estimate of the probability of a successful (lethal) engagement of the target by that weapon.
Weapon Selection Heuristics
These rules are combined to prioritize weapon selection for target engagement. All rules are evaluated; the order listed is not significant. Once the ideal weapon is selected to engage a particular target, the message is routed to the weapon across the network.
- Prefer weapons with highest percentage of remaining ammunition
- Prefer weapons with highest probability to hit this target
- Prefer weapons with highest probability to kill this target
- Prefer weapons with shortest engagement time (flight time) to this target
- Prefer weapons with shortest delay until they are ready to fire (due to reloading, recharging, etc)
Because damaged networks can isolate sensor and weapon nodes, the network may select a sub-optimal weapon system for engagement if the best available system cannot be reached through the network.
Emergent Behavior
Several emergent behaviors manifest as a result of the federated nature of the network. Weapon engagements are distributed among targets, which maximizes the number of simultaneous engagements. Multiple weapons do not engage the same target at the same time (although they may engage sequentially, provided the preceding engagement was unsuccessful).
Large networks react to damage. As long as any sensor can talk to any weapon, the network is a functioning defense system. When the overall network is split (due to damage) into multiple smaller networks, each smaller network that contains at least one sensor and one weapon will function as a defensive system. This means that split networks are more likely to double up weapon engagements against a single threat. This is a desirable behavior as a damaged network will need to concentrate its fire in response to threats penetrating the defense boundary.
Weapons
Weapon engagements are indicated by black lines. The line persists for the duration of an engagement, including flight time and impact. Hostile engagements are indicated by red lines that persist for the duration of the engagement. Data parameters such as flight time, ammunition, probability-to-hit, and probability-to-kill can be used to simulate different weapon platforms.
Sensors
Sensors can simulate surface (land and sea), air, and space-based systems such as surface radar, AWACs aircraft, and satellites. Sensors are the primary driver of the SoS, providing information messages to the network to drive weapon system responses. Data parameters such as scan frequency, range, and probability-to-detect can be used to simulate different sensor platforms.
Hostiles
Hostile aircraft select targets from among sensors, weapons, and cities. Sensors and weapons can be destroyed by hostile attacks, removing their functions from the network, and eliminating their node from the network topology. Cities accumulate damage, indicated with small red numbers next to their icons (yellow). The numbers represent ~10's of millions of dollars in damage and serve a surrogate for casualties when comparing runs. The implicit assumption is that property damage magnitude and casualty count values are correlated.
Hostiles are set to attack target types based on the hostile type. Small fast hostiles are set to attack defense network systems such as sensors and weapons (tactical). Large slow hostiles are set to attack urban centers (strategic). Hostiles randomly select their target when created, move to the target, attack until their stores are empty, then egress from the area in a random direction.
Analysis
Simulation trials are analyzed by comparing a number of statistics, such as the total cost of the simulation (cost of expended ammunition, price of weapon/sensor nodes, cost of network, and cost of damage inflicted by the enemy). Additional statistics include the total number of enemies neutralized by a given network, and the duration from initial enemy contact until the first successful enemy attack (an attack that destroys a weapon / sensor node or damages a city).
During development, analysis is conducted using hot-key controlled functions to configure and launch multi-trial simulation runs and track performance metrics. This demo does not have a web-interface to access those analysis tools. The simulation can be configured via the existing web interface (below) and run with multiple hostile aircraft in order to observe engagements and the effects of network degradation.