To characterize the performance of Paradyne's Hotwire MVL System, we simulated three real-world digital subscriber line (DSL) loops we called the good, the bad and the ugly using a Consultronics DSL loop simulator and a Consultronics noise injector.
For the good loop, we simulated a CSA Loop 4, which is a 6,800-foot loop with
an 800-foot bridge tap at the customer premises and a 400-foot bridge tap at the central office (CO). Loop 4 consists of standard 26 American Wire Gauge (AWG) wire. Each bridge tap is a branch point on a loop. This loop represents 60% customer coverage, meaning that 60% of customer loop conditions are equal to or better than this scenario. We injected a composite noise of 140 dBm/Hz background noise. The human ear can barely detect this noise level,
which ANSI says is representative of high-frequency noise on adequately performing subscriber loops. Because we only had one noise source, we placed it at the receiving end of the traffic for a unidirectional traffic flow and at the CO side of the loop for a bi-directional traffic flow.
The bad loop represents 90% customer coverage. For the bad loop, we adopted a modified ANSI Loop 7, which consists of a 13,500-foot loop with 26 AWG wire and a 1,000-foot 26 AWG bridge tap at the customer premises. We again injected a composite background noise of -140dBm/Hz; and we further increased interference by adding four HDSL, four ISDN and 24 T-1 pairs in an adjacent binder on the loop.
For the ugly loop we simulated a 24,000-foot twisted-pair loop with 24 AWG wire, which represents 99% customer coverage. We injected the same background noise and signal disturbers on this loop as we injected on the bad loop.
For each loop condition, we tested single DSL link performance, on-hook/off-hook transition times and the performance of two modems over a single DSL link.
For single DSL link performance, we designed a five-test suite that measured maximum throughput without loss from the modem through the DSL Access Multiplexer (upstream); maximum throughput without loss from the DSLAM through the modem (downstream); bidirectional throughput without loss; average latency from the modem through the DSLAM (upstream); and average latency from the DSLAM through the modem (downstream). We repeated each test for 64-, 512-, 1,024-, and 1,518-byte packets, and we repeated the test suite over the good loop while the phone was off-hook.
We measured latency using a 100K bit/sec traffic stream for the good and bad loops, and a 20K bit/sec traffic stream for the ugly loop.
Finally, we tested the amount of time it takes to stabilize data throughput when going between on-hook and off-hook states. We also ran a dual-modem test and measured bidirectional throughput without loss to see how the system performs with two modems on one DSL link.
RELATED LINKS Back to the main review
