Glossary
Terms on the wire
CRC, QP, CIC, and the rest — short definitions for lab work.
34 / 34 terms
- CRC
Cyclic Redundancy Check — a polynomial checksum over frame bits used to detect (not correct) transmission errors. Variants differ by width, poly, init, and reflection (e.g. CRC-15 CAN, CRC-16 Modbus, CRC-32 Ethernet FCS).
- FCS
Frame Check Sequence — the trailer CRC on a data-link frame (Ethernet’s 32-bit FCS). Computed over the addresses through payload; bad FCS frames are dropped by the MAC.
- NRZ
Non-Return-to-Zero — a line coding where bit values map directly to two voltage levels without returning to zero between bits. UART and CAN use NRZ with additional framing or stuffing rules.
- Baud
Symbol rate: signaling elements per second. For binary NRZ UART, baud equals bit rate; for multi-level or encoded PHYs, bit rate can exceed baud.
- PDU
Protocol Data Unit — the unit exchanged at a given layer (frame, packet, segment, message), including that layer’s headers/trailers around its payload.
- SDU
Service Data Unit — the payload a layer is asked to carry for the layer above. An upper-layer PDU becomes the lower layer’s SDU.
- MTU
Maximum Transmission Unit — largest payload a link layer will accept (classic Ethernet 1500 bytes). Path MTU is the minimum MTU along a route; discovery failures cause black holes when DF is set.
- QP (Queue Pair)
RDMA endpoint consisting of a send queue and a receive queue. Work requests posted to the QP are completed on a completion queue; QP state must reach RTS before traffic.
- MR (Memory Region)
Registered virtual memory for RDMA. Registration pins pages and yields local/remote keys (lkey/rkey) that authorize DMA to those buffers.
- CQ (Completion Queue)
RDMA queue where the NIC reports completed work requests. Polling or event-driven consumers drain CQEs; overruns raise asynchronous errors.
- B-channel
ISDN Bearer channel — a 64 kb/s clear digital channel for voice or data. BRI has 2B; PRI has 23B (T1) or 30B (E1).
- D-channel
ISDN Delta channel — carries LAPD/Q.931 signaling. BRI D is 16 kb/s; PRI D is 64 kb/s. Without D-channel link, B-channels cannot be controlled.
- RAS
Registration, Admission, and Status — H.225.0 gatekeeper protocol (usually UDP) for endpoint registration, call admission (ARQ/ACF), and bandwidth status.
- Bit stuffing
Insertion of a complementary bit after a run of identical bits (CAN: after five) so receivers can maintain sync. Stuff errors indicate lost edges or wrong bitrate.
- Bus arbitration
Method for deciding which transmitter owns a shared medium. CAN uses non-destructive bitwise CSMA/CR where the lowest ID wins; Ethernet uses CSMA/CD or switched full-duplex.
- ACK
Acknowledgement — indication that data was received. Appears as CAN ACK slots, I²C ACK bits, TCP ACK flags, or SIP/ response codes depending on the stack.
- PCI (ISO-TP)
Protocol Control Information in ISO-TP — the first data byte(s) that mark Single/First/Consecutive/Flow-Control frames and carry length or sequence fields.
- DID
Data Identifier — 16-bit UDS key for Read/WriteDataByIdentifier. Layout and scaling of the returned bytes are OEM-defined.
- NRC
Negative Response Code — UDS byte after 0x7F explaining why a request failed (e.g. 0x33 securityAccessDenied, 0x78 responsePending).
- Offer/Answer
SDP negotiation model (RFC 3264): one side offers media/codecs; the other answers with the intersection. Used by SIP, WebRTC, and similar session protocols.
- SSRC
Synchronization Source — 32-bit RTP identifier for a single media stream clock. Receivers demux jitter buffers per SSRC; collisions should trigger remapping.
- Point code
SS7 network address for a signaling point. ITU uses 14-bit codes; ANSI uses 24-bit. OPC/DPC in the MTP3 routing label steer MSUs.
- CIC
Circuit Identification Code — ISUP field selecting which trunk timeslot between two exchanges a call uses, together with the OPC/DPC pair.
- ECMP
Equal-Cost Multi-Path — routing/switching technique that spreads flows across parallel paths using a hash (often 5-tuple). RDMA/RoCE needs port entropy to avoid polarization.
- PFC
Priority Flow Control (802.1Qbb) — per-priority pause on Ethernet to build lossless classes for RoCE. Misuse causes head-of-line blocking or pause deadlocks.
- ECN
Explicit Congestion Notification — IP/TCP (and RoCE DCQCN) marks packets instead of dropping them when queues build, allowing senders to cut rate before loss.
- GID
Global Identifier — 128-bit InfiniBand address (often IPv6-mapped). Used with GRH for routed IB; RoCEv2 maps GIDs to IP addresses on the RNIC.
- LID
Local Identifier — 16-bit InfiniBand address assigned by the subnet manager for forwarding inside a single subnet.
- LNP
Local Number Portability — keeps an E.164 number when the subscriber changes carriers. Originating networks dip an NPDB and route on the returned LRN rather than the dialed NPA-NXX alone.
- LRN
Location Routing Number — the routing number returned by an LNP dip that identifies the switch currently serving a ported telephone number.
- CNAM
Calling Name Delivery — database-backed display name for the calling party, historically queried over SS7/LIDB and increasingly mapped into SIP identity headers or post-STIR dips.
- LIDB
Line Information Database — SS7-reachable database of line attributes used for collect/third-number billing validation and often CNAM name records.
- AIN
Advanced Intelligent Network — North American SSP/SCP trigger model (Info_Analyzed and related) used for LNP, toll-free, screening, and number translation over TCAP.
- NPAC
Number Portability Administration Center — regional clearinghouse that coordinates LNP port orders and broadcasts updates to carrier SOA/LSMS → NPDB systems.