When talking about Denial of Service attacks one generally discusses attacks to the CPU, RAM, or memory on machines. I have yet to hear of an attack going after the cables which are used to transport the data (so by sending more data through certain cables than they can handle), why is this? Are these cables (either optic, CAT5 etc...) sufficiently robust that it is infeasible?
If I have physical access to a medium I can more simply deny the use of the medium by jamming rather than trying to overwhelm the endpoints with valid requests. If I have a cable tap I can just spew electronic noise into the wire. If I have a wifi transceiver I can just spew electronic noise into the air. You don't need to generate properly formatted transmissions to overwhelm the service.
If you exceed the throughput of a cable, the result is not "more data in the cable than it can handle". The signalling scheme defines a bitrate, there's never larger amounts of data sent using this signalling scheme. Instead, the transmit queues of the sender will grow, because data is entering the queue faster than it is leaving.
If the queue can grow unbounded, this becomes an attack on the RAM of the sender. If not, data is discarded rather than going into the cable.
You can't convince standard hardware to overdrive the signals going onto its cable. First, the transmission electronics aren't capable of generating voltages that would damage the cable. Secondly, additional data doesn't result in greater voltages.
It's certainly possible to do a denial of service on a network link; this is part of what happens with a Distributed Denial of Service (DDOS) attack. However, in many cases it's easier (cheaper) to prevent network saturation than it is to prevent the exhaustion of other resources.