Not really a simple task, you will probably need a lot of network reconnaissance experience to be able to do this effectively.
If the IDS performs active response (where connections are reset in response to detected intrusions) you can identify the presence of the IDS by the packets it sends in order to terminate a session, which from memory are TCP RST packets for TCP connections and ICMP Error Port Unreachable for UDP connections.
The the IDS is running inline in IPS mode, you can identify its presence by something like traceroute, if firewall rules cooperate and the IP is routing the traffic as opposed to transparently bridging it. You probably wont be able to determine just from the appearance of the box in a traceroute that it is a IPS however.
Otherwise, you could identify the presence of a box running an IDS by standard discovery techniques (ping scans, port scans, etc - again dependant on firewall rule configuration), although again you may not immediately be able to determine that the host is an IDS based on this information alone.
Detecting a properly configured passive IDS sensor would be difficult, because they are generally setup to not have an IP address bound to the interface which is monitoring traffic. In this case, alerts and configuration changes would be sent via a seperate interface, which might be connected to a seperate network (perhaps to a network designated for monitoring only).
Hope that helped.