Write the 4.2.8 decision a half-duplex MAC makes after each collision: wait and retry, or give up, and why.
Hardware context: An Ethernet MAC and physical coding sublayer: frame builder and checker, reconciliation sublayer and block encoder. Build, check and pace Ethernet frames, and code them for the physical layer as IEEE 802.3 specifies.
After a collision and its jam, a half-duplex MAC decides what to do with the frame. It may wait a random number of slots and retry. Or it gives up and reports why.
Write retry_decision. It gets the collision count, the packet bit where the collision was seen, a random word, the rate and one option bit. It returns a verdict and the wait.
Take the parameter values from Table 4-2 in the lesson: slotTime, attemptLimit and backoffLimit at each rate.
Signals in the Timeline
| Signal | Width | Role |
|---|---|---|
test_number | 8 | input of each check |
Full specification
function automatic [31:0] retry_decision(input int n, input int at_bit, input [31:0] rnd,
input int mbps, input bit stop_late);
// returns {verdict[1:0], delay_bit_times[29:0]}
| Rule | Value |
|---|---|
| Late collision | seen after the first slotTime bits of the packet, preamble included. 4.2.8 counts from the end of the 64 header bits, which is the same point |
| Late, at 1000 Mb/s | the MAC must stop: verdict 2 |
| Late, at 10 or 100 Mb/s | retried like any other collision. With stop_late = 1 the MAC takes the option 4.2.8 allows and stops: verdict 2 |
| Too many collisions | when n reaches attemptLimit: verdict 1 |
| Both at once | report the late collision. 4.2.8 does so at 1000 Mb/s; at 10 and 100 Mb/s this problem keeps the same order |
| Otherwise | verdict 0: wait r × slotTime bit times, 0 ≤ r < 2^k, where k is n but never more than backoffLimit |
| r | the low k bits of rnd. This is a bench rule: the standard asks for a uniform random r |
| Giving up | the delay bits are 0 |
Examples
Example 1
First Retry at 10 Mb/s- Input
n = 1, at_bit = 200, rnd = FFFFFFFFh
- Output
delay_bit_times = 512, verdict = retry (0)
Bit 200 is inside the first 512-bit slot, so this is an ordinary collision. k = 1, so r is bit 0 of rnd: 1. The wait is one slot, 512 bit times.
Example 2
Third Retry- Input
n = 3, at_bit = 300, rnd = 12345675h, 100 Mb/s
- Output
delay_bit_times = 2560, verdict = retry (0)
k = 3, so r is the low three bits: 5. The wait is 5 × 512 = 2,560 bit times.
Example 3
A Late Collision at 1 Gb/s- Input
n = 1, at_bit = 5000, rnd = 3, 1000 Mb/s
- Output
delay_bit_times = 0, verdict = give up: late collision (2)
At 1000 Mb/s one slot is 4096 bits, and bit 5000 is past it. The MAC must stop: verdict 2, delay 0.
Constraints and Contract
- Keep the name and the arguments.
- n is at least 1, at_bit at least 1. mbps is 10, 100 or 1000.
- stop_late matters only at 10 and 100 Mb/s.
- Off limits: the names
model,rng,tb,rootand the modules of the bench.
Hints
Hint 1
How many bits is one slot at 1000 Mb/s? Is a collision on that very bit late?
Hint 2
At 100 Mb/s, what does a late collision change when stop_late is 0?
Hint 3
What happens to the range after the 11th collision, and after the 16th?
What the grader checks: The fixture checks 17 literal cases against an independent JavaScript model, including every window edge and the 16th collision. It checks 800 random cases against a private model. Taking r from the low bits of rnd is a bench rule; the standard asks for a uniform random r. 4.2.8 does not fix the status a 10 or 100 Mb/s MAC reports when it takes the option. This problem uses the late-collision verdict.
19 checks on Icarus Verilog · server. Run the examples first, then check the full regression. Full checks use the server judge. Downloads contain your source only; samples require the simulation service. A passing result establishes the tested cases, not a proof for all possible inputs.
See all 2 steps and the final capstone