**IMPACT:** Potential concern from Motorla Solutions, as IPSC is a proprietary
**METHOD:** Reverse engineering by pattern matching and process of elimination
**PROPERTY:**
This work represents the author's interpretation of the Motorola(tm) MOTOTRBO(tm) IPSC protocol. It is intended for academic purposes and not for commercial gain. It is not guaranteed to work, or be useful in any way, though it is intended to help IPSC users better understand, and thus maintain and operate, IPSC networks. This work is not affiliated with Motorola Solutions(tm), Inc. in any way. Motorola, Motorola Solutions, MOTOTRBO, ISPC and other terms in this document are registered trademarks of Motorola Solutions, Inc. Other registered trademark terms may be used. These are owned and held by their respective owners.
**PRE-REQUISITE KNOWLEDGE:**
This document assumes the reader is familiar with the concepts presented in the Motorola Solutions(tm), Inc. MOTOTRBO(tm) Systems Planner.
When communications exchanges are described, the symbols "->" and "<-"areusedtodenotethe*direction*ofthecommuncation.Forexample,"PEER-> MASTER" indicates communcation from the peer to the master. For each exchange outlined, the initiator of the particular communication will be on the left for the duration of the particular item being illustrated.
The IPSC system contains, essentially, two types of nodes: Master and Peer. Each IPSC network has exactly one master device and zero or more peers, recommended not to exceed 15. IPSC nodes may be a number of types of systems, such as repeaters, dispatch consoles, application software, etc. For example, the Motorola RDAC application acts as a peer in the IPSC network, though it doesn't operate as a repeater. The IPSC protocol supports many possible node types, and only a few have been identified. This document currently only explores repeaters - both Master and Peer, and their roles in the IPSC network.
All IPSC communication is via UDP, and only the master needs a static IP address. Masters will operate behind NATs. A single UDP port, specified in programming the IPSC master device must be mapped through any NAT/stateful firewalls for the master, while peers require no special treatment.
All nodes in an IPSC network maintain communication with each other at all times. The role of the master is merely to coordinate the joining of new nodes to the IPSC network. A functional IPSC network will continue without its master, as long as no new nodes need to join (or existing nodes need to re-join after a communications outage, etc.) This is one of the most important core concepts in IPSC, as it is central to the NAT traversal AND tracking of active peers.
Each peer will send keep-alives to each other peer in the IPSC network at an interval specified in the devices "firewall open timer". The elegantly simple, yet effective approach of IPSC, uses this keep-alive to both open, and keep open stateful firewall and NAT translations between peers. Since each device handles all communications from a single UDP port, when a device sends a keep-alive or a registration request to another device, the source-destination address/port tuple for that communication is opened through stateful devices. The only requirement to maintain communication is that this timer be shorter than the UDP session timeout of network control elements (firewalls, packet shapers, NATs, etc.) Moreover, it does NOT appear that all devices in the IPSC network require the same setting for this. Each device would appear to maintain its own set timing without interference from different interval settings on other nodes in the IPSC.
The following sections of this document will include various packet types. This is a list of currently known types and their meanings. Note: The names are arbitrarily chosen with the intention of being descriptive, and each is defined by what they've been "observed" to do in the wild.
Most IPSC networks will be operated as "authenticated". This means that a key is used to create a digest of the packets exchanged in order to authenticate them. Each node in the IPSC network must have the authentication key programmed in order for the mechanism to work. The process is based on the SHA-1 digest protocol, where the "key" is a 20 byte hexadecimal *string* (if a shorter key is programmed, leading zeros are used to create a 20 byte key). The IPSC payload and the key are used to create the digest, of which only the most significant 10 bytes are used (the last 10 are truncated). This digest is appended to the end of the IPSC payload before transmission. An example is illustrated below:
The IPSC network truly "forms" when the first peer registers with the master. All peers register with the master in the same way, with a slight variation from the first peer. Below is a descirption of the process and states in creating a connection, as a peer, and maitaining it.
There are various states, timers and counters associated with each. When peers or the master send us requests, we should answer them immediatley. Our own communcation with them is timed, and may share the same timer. Counter values should be the same for every master and peer in an IPSC. They don't have to be, but that is what mother M does, and it saves a lot of resources.
The following illustrates the communication that a peer (us, for example) has with the master. The peer must register, then send keep-alives at an arbitrary interval (usually 5 - 30 seconds). If more than some arbitrary number of keep-alives are missed, we should return to the beginning and attempt to register again -- but do NOT elimiate the peers list, as peers may still be active. The only additional communcation with the master is if the master sends an unsolicited peer list. In this case, we should update our peer list as appropriate and continue.
Once we have registered with the master, it will send a peer list update to any existing peers. Those peers will **immediately** respond by sending peer registrations to us, and then keep alives once we answer. We should send responses to any such requests as long as we have the peer in our own peer list -- which means we may miss one while waiting for receipt of our own peer list from the master. Even though we receive registration requests and keep-alives from the peers, we should send the same to them, even though this is redundant, it is how we ensure that firewall UDP sessions remain open. A bit wonky, but elegant. For example, a peer may not have a firewall, so it only sends keep-alives every 30 seconds, but we may need to every 5; which we achieve by sending our own keep-alives based on our own timer. The diagram only shows the action for the *initial* peer list reply from the master. Unsolicited peer lists from the master should update the list, and take appropriate action: De-register peers not in the new list, or begin registration for new peers.
