Short for either constant or Latin celeritas, meaning “speed”. The speed of light. In a vacuum, c = 299,792,458 m/s. This is often denoted c0, since it’s now known that the speed varies with the medium. It appears that photons crossing great (galactic) distances travel slightly slower than c0. Researchers suggest this is due to a rare quantum-mechanical phenomenon called vacuum polarization, in which the photon spontaneously forms an electron-positron pair that almost immediately recombines to form a new photon with the same vector.
Command, Control, Communications, & Intelligence. A military term for the systems by which commanders manage operations. Sometimes extended to C3ISR (C3I Surveillance & Reconnaissance).
One of a family of stable explosives. It’s about 91% RDX (a.k.a. cyclonite or cyclotrimethylene trinitramine), the rest of the weight being plastic binder (polyisobutylene), plasticizer, and, nowadays, a taggant to identify it.
Command, Control, Communications, Computers, & Intelligence. An update of C3I to reflect the importance of computers. Sometimes extended to C4ISR (C4I Surveillance & Reconnaissance).
Cable connectors used in electronics and RF systems are broadly divided into four categories by size: standard (including the UHF and type N connectors), miniature (BNC, TNC, SHV, MHV, and mini-UHF), sub-miniature (SMA, SMB, SMC, SMK, SMP, MCX, APC-2.4, APC-3.5, K, and BMA), and micro-miniature (SSMA, SSMB, SSMC, SSMP, SBMA, and MMCX, all of which are scaled-down versions of sub-miniature connectors). “RP-nnn” or “R-nnn” means a reverse-polarity version of a standard connector, using a male contact in a female connector housing and vice-versa. There are RP-BNC, RP-MMCX, RP-N type, RP-SMA, and RP-TNC connectors, and probably others. Also, see DC.
These are for carrying data signals in electronics and RF systems. For telephone, network, fiber-optic, and computer video cables, see USOC, 802.3, fiber, and graphics, respectively.
1.85mm – Very similar to the SMA and APC-3.5, but not compatible.
APC-2.4 – Amphenol Precision Connector-2.4mm. A threaded 50Ω microwave (to 50 GHz) connector, also called an OS-50.
APC-3.5 – Amphenol Precision Connector-3.5mm. Originally from HP, this threaded 50Ω microwave (to 34 GHz) connector has very low VSWR and can mate with SMA connectors.
APC-7 – Amphenol Precision Connector-7mm. Also originated at HP, this 50Ω connector has exceptionally low VSWR up to about 18 GHz, and is hermaphroditic (i.e. any APC-7 can mate with any other).
APC-N – Amphenol Precision Connector type N. A 50Ω, low-VSWR microwave (to 18 GHz) connector with 11/16"-24 threaded coupling. It mates with N type connectors.
BMA – Blind-Mate. A smooth, cylindrical, snap-coupling, bayonet-to-socket connector with 50Ω characteristic impedance. The plug (male) diameter is 0.210". Used up to 22 GHz. The connectors can be mated without visual checking to orient them, hence the name.
BMA, male (left) and female (right)
BMC – Blind-Mate Connector. Same thing as BMA.
BNC – Bayonet (originally Baby) Neill-Concelman. This very common coaxial cable connector is about ½" diameter, uses a pin-to-socket mechanism and quarter-turn locking lugs rather than threaded coupling, and is manufactured with 50Ω or 75Ω characteristic impedance. It’s rarely used above 1 GHz, although 50Ω versions are adequate to about 4 GHz, and 75Ω ones to 12 GHz or more. As for the name, “bayonet” describes the coupling mechanism, and American engineers Paul Neill (1882-1968) and Carl Concelman (1912-1975) were the creators of the N and C connectors, respectively, from which they derived the BNC at Bell Labs in the late 1940s. Because it’s smaller than its ancestors, it was originally dubbed a “baby NC” connector, but this became “bayonet NC” after the later development of the TNC. Some sources misinterpret the acronym to mean bayonet nut connector, bayonet navy connector, British naval connector, and other, less probable things.
BNC, male (left) and female (right)
BNT – Similar to the BNC, but used with triaxial cable.
C type – Named for its inventor, Carl Concelman of Amphenol. This unthreaded coaxial connector is based on the N type connector, but uses bayonet-to-socket coupling and locking lugs to overcome the small inductance created by the center pin. The male is ¾" diameter.
F type – A coaxial connector with 3/8"-32 threaded coupling, commonly used for TV antenna and cable feeds. The conducting pin of the male connector extends about ¼" beyond the socket.
F type female and male cable connector
G type – A snap-on coaxial connector for signals up to 3 GHz.
IDC – IBM Data Connector. Hermaphroditic. Used in token ring networks, hence rare.
K type – A 2.92mm connector.
MCX (or µCX) – Micro-CoaX. A 1980s European snap-on coaxial connector, similar to the SMB but 30% smaller. Versions with a 90° bend can swivel freely around the connection point. The 50Ω version operates up to 6 GHz, and the 75Ω to 10 GHz. It’s used in GPS and automotive equipment, cellular phones, and telemetry.
MCX, female (left) and male (right)
mini-UHF – A coaxial connector designed for cellular phone systems. It has 3/8"-24 threaded coupling and 50Ω characteristic impedance, and is used for signals up to 2.5 GHz. It’s based on the larger UHF connector.
MMCX – Micro-Miniature CoaX. An even smaller version of the MCX, conforming to European CECC 22000 specifications.
N type – Named for its inventor, Paul Neill of Bell Labs, this was the first true microwave coaxial connector. It was created in 1942 for use in radar systems. It has 5/8"-24 threaded coupling and 50Ω or 75Ω characteristic impedance. Connectors with different impedances won’t mate. It’s commonly used for LANs and various communications & test equipment with signals up to 11 GHz and 1500V peak, but there are precision versions that operate up to 18 GHz.
Type N, male (left) and female (right)
QMA – Quick-lock Miniature type A. A snap-coupling variant of the SMA connector.
QN – Quick-lock type N. A snap-coupling variant of the type N connector. The 90° angle plug swivels.
RCA – Radio Corporation of America. An analog audio/video connector with a fat central pin and flanges around the rim of the plug. Audio and video signals require separate connections.
RCA, male (left) and female (right)
SMA – Sub-Miniature type A. A ¼"-36 threaded coaxial connector developed in the 1960s for use with 0.141" semi-rigid cable (RG-402), and sometimes used as a simplex fiber-optic connector. It has 50Ω or 75Ω characteristic impedance. Precision types work up to 26.5 GHz. It’s used in applications that require small size and high frequency. There are sub-types SMA-905 and -906.
SMA, male (left) and female (right)
SMB – Sub-Miniature type B. A snap-coupling (rather than threaded) coaxial connector, slightly smaller than the SMA. It’s common in on-board RF designs. The 50Ω SMB connector is rated to 4 GHz, and the 75Ω connector to 2 GHz. Connectors with different impedances might or might not mate.
SMB, female (left) and male (right)
SMC – Sub-Miniature type C. A coaxial connector with 3/16"-32 (50Ω) or 5/16"-32 (75Ω) threaded coupling. The 50Ω connector operates up to 10 GHz. It’s used in microwave telephony and non-military communications where small size and resistance to vibration are important.
SMK (or 2.9mm) – Sub-Miniature type K. Very similar to the SMA and APC-3.5, with which it will mate.
SMP – Sub-Miniature type P. Tiny snap-coupling connector.
TNC – Threaded Neill-Concelman. A variant of the BNC connector using 7/16"-28 threaded coupling rather than locking lugs. It was developed in the late 1950s because of the noise generated by BNC connectors under extreme vibration. It has 50Ω or 75Ω characteristic impedance, operates up to 11 GHz, and is used in military and aerospace applications where vibration is a concern.
TNC cable connectors
TNT – Based on the TNC, but for triaxial cable, which has three conductors rather than two. It isn’t compatible with the TNC.
triax – Based on the BNC, but, again, for triaxial cable, and not compatible with the BNC. It often has three locking lugs rather than the BNC’s two.
Triax female cable connector
twinax – The same size as type N, but for use with two-conductor cable. See twinax.
UHF – Ultra-High Frequency. A coaxial connector developed by E. Clark Quackenbush (!) at Amphenol in the 1930s, for radio applications. (At the time, “UHF” referred to the radio frequency range now called VHF.) The plug version is commonly called the PL-259, its military part number. The UHF has 5/8"-24 threaded coupling, with a large, rounded pin in the male connector and a serrated rim on the female. It doesn’t have a constant impedance, which limits its use to signals under 300 MHz. It can handle up to 500V peak, and is used in low-frequency communication equipment such as CB radios and public address systems.
UHF, male (left) and female (right)
ZMA – Zone Multicast Access, maybe. New (~2002). Essentially a 50Ω SMA that uses a bayonet-coupling, quarter-turn locking lug connection like the BNC. Unlike the BNC, it’s made to tolerate vibration. Inter-compatible types exist rated for 3 GHz (high-power), 12.4 GHz, or 18 GHz.
Provides access to TV signals, and optionally broadband Internet connectivity, over coaxial cable from a cable TV company. Most of them are (2015) based on a physical HFC network, and use DOCSIS signaling to carry Internet traffic. Warning: Even if a modem supports the DOCSIS version the provider uses, there’s no guarantee it will be compatible with their system. Compare DSL.
In 2001, Road Runner and @Home were the two biggest cable modem ISPs. By 2011, it was Comcast.
In computing hardware, a small, high-speed repository for information that a CPU uses often. There are instruction caches and the more common data caches. For example, modern hard drives have an integrated data cache called a disk cache. Most references to cache, however, are talking about the ones that speed up the CPU’s use of main memory (DRAM). DRAM technology hasn’t been able to keep pace with the ever-growing bandwidth of processors. To compensate, CPU designers have developed a multi-layered system of memory caches called level 1 (L1), level 2 (L2), etc. They mainly use SRAM, which is inherently faster than DRAM, but also relatively bulky and expensive. Virtually all modern processors have L1; the others aren’t universal.
L1 cache is on the CPU itself, running at core speed. It’s divided into a separate instruction cache and data cache. L1 caches return hits – that is, have the cached data that the CPU needs – about 95% of the time.
L2 cache is larger and somewhat slower than L1, and is strictly a data cache to backstop L1 data cache misses. It started out on the motherboard, with access via the I/O bus or system bus. In 1998, Intel introduced the back side bus or cache bus as a separate, dedicated connection to the L2, along with new, proprietary chip module sockets (Slot 1 & Slot 2) that accommodated it. This led to putting the L2 into a multi-chip module with the CPU, or on the CPU die proper, where it resides in modern designs.
L3 cache was introduced around 2000. Whereas each CPU core has its own dedicated L1 and L2, the L3 shares data between the cores of multi-core processors, and supports sharing of on-chip resources such as MXT.
L4 cache is a shared memory resource on multi-chip modules and multi-core processors. It’s bigger and slower than L3, and is usually for integrated graphics.
Caches can be direct-mapped, meaning each block of cache contains data from one specific block of main memory; fully associative, meaning main memory data can be stored anywhere in cache; or n-way associative, meaning data from each block of main memory can be stored in any of n cache blocks. These approaches trade off cache hit rate against cache search speed.
Cache management has traditionally been entirely in hardware. Researchers have proposed (2013) that the OS kernel could do it much more efficiently. This would require the CPU to have a hardware cache monitor for the OS to use, but could dramatically improve system speed and power consumption.
Computer-Aided Manufacturing. This process produces data files used directly in the manufacture of printed circuits. Types of CAM files are the Gerber file, which controls a photoplotter; the NC Drill file, which controls a drilling machine; and the fab and assembly drawings.
Content-Addressable Memory. A sort of inverse RAM: it accepts a data input, and provides as output the address or range of addresses at which that data is stored. It’s used in systems that need to do high-speed search, such as network hardware and PLDs. Standard CAM is binary, dealing in ones and zeros. Ternary CAM (TCAM) permits the system to apply a mask to the data sequence; the search ignores the state of any masked bits.
Compression Attached Memory Model. A JEDEC standard for DRAM in laptop and notebook PCs, originated by Dell, and expected to replace the thicker SO-DIMM starting in 2024.
Controller Area Network. A half-duplex serial data bus created to link microcontrollers in automotive applications, including OBD-II, and also used in aviation and industrial applications. It uses a twisted-pair line with 120 Ω characteristic impedance, runs at up to 1 Mb/s, uses CSMA/CD, and has a practical limit of 110 nodes. Each CAN message consists of a 1-bit start-of-frame, unique 11-bit (version 2.0A) or 29-bit (version 2.0B) node ID, 4-bit data length, up to 8 data bytes, 15-bit CRC, 7-bit end-of-frame, and a few other overhead bits. The CAN with Flexible Data Rate (CAN FD) extension, with a slightly different frame structure and longer CRC, supports data lengths up to 64 bytes and speeds greater than 1 Mb/s. Compare I2C, LIN, SPI.
Two conducting elements separated by a dielectric. Capacitance (C), the capacity to store electrical charge, is defined as the ratio of the charge stored to the electrical potential difference: C = Q/V. The SI unit of capacitance is the farad, or coulomb/volt. CV, or capacitance × voltage, is the standard measure of a capacitor’s volumetric efficiency. For a simple capacitor consisting of two parallel plates, capacitance depends on the area (A) of the conducting surface, the permittivity (ε) of the dielectric, and the distance (d) of separation: C = [(8.85 ε A) / (d ε0)] pF, where ε0 is the permittivity of free space, 8.854 × 10-12 F/m. The ratio ε/ε0 is the dielectric constant, k, of the material. For air, k = 1.
Schematic symbol, and examples of unpolarized and polarized capacitors
An ideal capacitor with capacitance C has an impedance Z defined entirely by capacitive reactance XC: Z = XC = 1/jωC, where ω is angular frequency of the voltage across it. Large capacitors, because they tend to have more internal (lead) inductance than smaller ones, can deviate significantly from this formula at high frequencies. In modern circuits, a capacitor’s equivalent series resistance (see ESR) is often more important than its capacitance. Common general types:
ceramic – Unpolarized. A ceramic dielectric between metal elements, in either disc (single-layer ceramic, or SLC) or monolithic (stacked multi-layer ceramic, or MLC) configuration. Ceramic caps are small and reliable, and have extremely low ESR. However, their capacitance varies significantly with temperature, and larger ones tend to have poor electrical properties. The dielectric type is indicated by an EIA code such as NP0/C0G, X7R, Y5V, or Z5U.
electrolytic – Polarized, although there are unpolarized compound designs. A thin oxide coating on the aluminum, niobium, or tantalum anode acts as the dielectric, and a dry (i.e. paste) or wet electrolyte as the cathode. Electrolytics can have much higher capacitance than the other designs, but also suffer high ESR and current leakage with less voltage and temperature tolerance, and perform poorly at higher frequencies (above 1 MHz). Wet-cathode types yield higher capacitance for a given volume, but are less robust, and fail as the electrolyte dries out. Tantalum-anode types are smaller, hardier, & pricier.
film – Unpolarized. Highly stable & reliable, modest capacitance ranges. There are two general sub-types. Thick-film, or film/foil, uses a plastic dielectric between metal foil and offers better precision and higher current handling. Thin-film, or metallized film, is metal vacuum-deposited on a dielectric substrate (e.g. silver-mica). It’s smaller, with lower losses and self-healing.
Electrolytic and film capacitors usually have the capacitance printed on them in picofarads (pF) or microfarads (µF). If you see MF (or mf, or mF) it means microfarads, not millifarads. A common notation, especially on ceramic disc capacitors, is a 3-number code based on picofarads. The first two numbers are the value, and the third is the number of zeros after it. For example, 474 means 470000 pF, which is 0.47 µF. The voltage tolerance and dielectric type might also be shown. See ultracap.
Channel Access Protocol code. A unique 7- or 8-digit code programmed into a pager. A CAP code is appended to the beginning of each page issued by a paging system, and a pager will accept only a page containing its own code. Groups of pagers can be programmed with an additional group CAP code so that they will all respond to the same message.
Carrierless Amplitude/Phase Modulation. An AT&T Paradyne signal encoding scheme for ADSL, with a suppressed carrier. It uses QAM – anywhere from 4 to 1024 symbols.
Completely Automated Public Turing Test to Tell Computers and Humans Apart. An automated challenge-response test that Web servers run to make remote users prove that they’re humans, not bots or automated scripts. For example, the site might present the user with one or more strings of distorted, camouflaged characters, sometimes actual words, which the user must type in correctly. More advanced tests than this are in use. Sophisticated bots sometimes pass, and sophisticated humans sometimes don’t. Compare DCG.
The CAPTCHAs for many Web sites are provided by a project named reCaptcha, which Google Books is using to crowdsource the transcription of scans of old text.
An old (17th century) type of catadioptric reflecting telescope. Hybrids of this design – the Schmidt-Cassegrain and Maksutov-Cassegrain – that compensate for its spherical aberration are much more popular.
Category #. A set of cables used in networks and phone systems. Most are defined by the standards EIA/TIA-568-A (1991-2001) and EIA/TIA-568-B (2001), usually shortened to T568A and T568B; the latter is now deprecated, but still in use. A Cat# cable contains one or more twisted pairs of 22 to 24 AWG wire, and, for networks, uses 8P8C modular connectors of the type inaccurately referred to as RJ45.
Cat1 – Not a formal standard. Basic telephone cable with one UTP, rated for signals up to 1 MHz. Not for networking. Normally has 6P2C modular connectors.
Cat2 – Not a formal standard. Telephone/network cable containing two UTPs. Used in token ring networks, rated to 4 MHz.
Cat3 – The original Ethernet (10BaseT) cable, containing four UTPs with 100 Ω characteristic impedance and 8P8C modular connectors. Rated to 16 MHz.
Cat4 – Similar to Cat3, but rated to 20 MHz. Used for faster token ring networks, and not much else.
Cat5 – For Fast Ethernet (100BaseT), four UTPs with 100 Ω characteristic impedance. Rated to 100 MHz.
Cat5e – E for Enhanced. Rated up to 100 MHz for Gigabit Ethernet (1000BaseT), although potentially able to run even 10GBaseT. Now considered the minimum acceptable grade.
Cat6 – (2002) Rated up to 250 MHz for 10GBaseT.
Cat6a – (2008) A for Augmented. Rated to 500 MHz, for 10 Gb/s Ethernet (10GBaseT). The ISO version of this standard restricts NEXT slightly more than the TIA/EIA version.
Cat7 – (2002) Not an ANSI standard; also called Class F. Rated to 600 MHz with improved EMI performance due to use of STP along with shielding of the cable bundle and connectors. For 10 Gb/s Ethernet.
Cat7a – To 1000 MHz.
T568A and T568B cables have four color-coded pairs of wires. For patch cables, T568A assigns wires to connector pins as shown in the figure below: 1 green-white, 2 green, 3 orange-white, 4 blue, 5 blue-white, 6 orange, 7 brown-white, 8 brown. T568B switches the orange pair and the green pair. Either works for a patch cable, as long as both ends are wired the same way. Wiring one end for T568A and the other for T568B makes a crossover cable.
Ethernet 8P8C patch cable with T568A wiring
The categories improve performance by such measures as shielding of individual pairs (using STP rather than UTP), shielding of the overall cable, thicker 23 or 22 AWG rather than standard 24 AWG wire, giving each wire pair in the cable a different pitch (number of twists per meter), and inserting cable splines to keep the pairs apart. Using solid rather than stranded wire also improves performance, but it’s not specified for any category because its stiffness is unsuitable for patch cables. Solid wire is only for fixed, in-wall cable runs.
Some cable is manufactured using copper-clad aluminum (CCA) rather than all-copper wire. CCA cables have higher resistance, are subject to oxidation of contacts, and do not meet TIA, NEC, and UL standards for Cat# cables.
The standards from Cat3 on are all compatible, but a network connection is a chain, and the weakest link in that chain sets the speed limit. There’s little point, for example, in buying Cat6 rated connectors for a Cat5 cable.
The terminal or electrode at which current leaves a system, hence where electrons enter. In a battery, fuel cell, or other DC source, the positive terminal is the cathode. In a circuit, it’s the negative terminal, i.e. the current sink. Compare anode.
A passive, linear RF band-pass filter. The cavity is half the length of the wavelength at the center of the passband. Multi-cavity filters with apertures between the cavities yield a wider passband and a sharper rolloff, and the bandwidth increases further with aperture size. Using circular rather than rectangular cavities reduces loss and bandwidth. Cavity filters can be mechanically tuned by changing the length.
Citizens Band radio. A RF band around 27 MHz set aside in a number of countries for short-range, PTT voice radio. The FCC defines it as 40 half-duplex channels with 10 kHz separation (and a few gaps) from 26.965 to 27.405 MHz. CB signals use standard AM at 4 watts or SSB modulation (either USB or LSB) at 12 watts, although illegal high-power transmitters exist.
Each channel can support one AM signal at a time, or one USB plus one LSB signal simultaneously. SSB conversations are supposed to use just channels 36-40. When CB radio manufacturers claim support for 120 CB channels, they’re ignoring these limitations and counting each channel three times, once per signal format.
Charge-Coupled Device. A semiconductor photosensor that produces a discrete charge corresponding to the intensity of incident light. CCDs with pixel interpolation were once the standard technology for digital cameras. CMOS-based cameras gained a foothold in the market by being less power-hungry and (initially) cheaper, and then caught up on resolution and image quality. CCDs still have better low-light resolution and greater dynamic range, giving them the edge in professional cameras.
Cold-Cathode Fluorescent Lamp. A type of fluorescent light common in electronics, especially as backlighting for LCDs. As the name implies, it generates an electron beam from the cathode of a vacuum tube without heating.
Complementary Code Keying. A new modulation format used in the IEEE 802.11b and 802.11g standards. CCK is a variation of M-ary orthogonal keying modulation, and uses I/Q modulation architecture with complex symbol structures. The spreading employs the same chipping rate and spectrum shape as the legacy 802.11 Barker word spreading functions, allowing for three non-interfering channels in the 2.4 GHz ISM band. See 802.11.
Code Composer Studio. A software development environment from Texas Instruments. It has compilers, libraries, and other tools specifically for TI DSPs and embedded processors, including a C-like programming language, GEL, used by debugger scripts. It replaced the old Code Composer, which was last updated in 2001.
Command & Control Switching System. A worldwide US military program that provides switching capabilities to the DRSN, for transferring classified data.
Closed Circuit Television. A TV system that transmits its signals electrically, by wire, rather than as radio waves over the air. Typical for security camera systems.
Compact Disc. A plastic disc medium for digital data. It’s normally 12 cm in diameter, but because the data track begins at the center and spirals outward, the disc can be physically smaller. Early CD drives varied the spin rate of the disc to keep the read speed constant. Current designs keep the spin rate constant and allow the read speed to vary, increasing as the read head moves farther from the center. The speed rating for a drive is given as 8x, 32x, 48x, etc., which refers to multiples of the read speed of the very first CD-ROM drives.
A typical disc holds about 650 (or 700) MB of data, roughly 74 (or 80) minutes of audio using 16-bit, 44100 Hz sampling of two stereo channels with no compression. CD-ROM refers to a drive with no write capability, or a mold-pressed disc that can’t be written to. CD-R (recordable) means a disc that can be laser-written only once (WORM), or a drive that can write only to this type of media. CD-RW (rewriteable) is a disc that can be overwritten, or a drive that supports multiple writes, sometimes even to a CD-R disc. CD-RW discs don’t work in all CD players. Discs must be properly formatted, for data or for audio, before writing. For audio discs, disc-at-once (DAO) writing puts all information on the disc at one pass in one or more tracks, and prevents further data being added later. Track-at-once (TAO) audio writing takes a little more space, but allows data to be written a track at a time on multiple occasions.
To clean a disc, use a soft, lint-free cloth and soapy water. Other approaches cause damage, even if it isn’t visible to the eye, and this includes commercial resurfacing products.
DVD drives can also read CDs, and combination drives are able to write both as well.
Cumulative Distribution Function. A function giving the probability that the random variable X is less than or equal to another variable x for all values of x. The CDF is found by integrating the pdf (for a continuous random variable X) or summing up the PDF (for a discrete random variable X).
Compact Disc File System. The 32-bit, protected-mode file system (and driver) for CDs on Windows 95 and later, replacing the slower 16-bit, real-mode MSCDEX.
Code Division Multiple Access. A.k.a. direct-sequence CDMA (DS-CDMA). Any of several DSSS multiplexing protocols pioneered by Qualcomm for jam-resistant military radios (see LPI) and adopted for digital mobile phones. A CDMA application might have, say, 6 signals sharing a 1.25 MHz band. Every CDMA base station has a correlation receiver that stores all of the spreading codes in use on the system, indexed by ESN.
Wideband CDMA (W-CDMA) uses a much faster spreading code, yielding a 5 MHz spectrum, and uses either FDD or TDD for its uplink and downlink. It’s specified to provide data rates of 114 kb/s mobile, 384 kb/s pedestrian, and 2 Mb/s stationary, but the real-world figures will be more like 64-114 kb/s.
CDMA is popular for 2G and later digital cellular phone systems. In the US, these have included N-CDMA and cdmaOne (IS-95). CDMA2000 (IS-2000) is a Qualcomm standard to help N-CDMA systems switch from 2G to 3G.
Content Delivery (or Distribution) Network. A network that maintains copies of a particular type of content – video, music, files, Web pages, social-network pages, etc. – on multiple servers for access by large numbers of users. If part of a network is overloaded or unavailable, other parts serve the content.
Cellular Digital Packet Data. Now obsolete. A wireless open standard created to provide the old analog AMPS with access to packet-switched networks, using IP and CLNP. It has been adapted to IS-95 and ?IS-146?. It uses idle network capacity caused by pauses in telephone conversations and gaps between calls to transmit two-way, 19.2 kb/s (or less) packet data. It’s gradually going away.
Clinical Decision Support. Refers to computerized systems that generate medical test or treatment recommendations from input. So far (2015), they are not reliable.
The ancient Greek model of a geocentric universe – mentioned by Plato, elaborated by Aristotle, and laid out in detail around 150 AD by Claudius Ptolemy – conceived a set of vast, concentric crystal spheres rotating around the Earth carrying the sun, moon, stars, and planets. This model endured, with many, increasingly tortured modifications, well into the Renaissance. Danish astronomer Tycho Brahe, after his 1577 observation of a comet that passed through the supposed spheres, was the first to mount a challenge to it. It still plays a role in astronomy, however – see equatorial system.
Cell Broadband Microprocessor, a new kind of processor chip developed for the Sony PlayStation 3 (PS3) game system, but with other uses intended. It has nine processors: eight specialized Synergistic Processor Elements (SPEs), each with just 256 kB of on-chip memory, supervised by one Power Processor Element (PPE) that doles out tasks and data. It’s very complex to code for.
A facility where subscriber lines (local loops), usually multiplexed through ONUs, meet the switching equipment that lets them access the telephone system and the Internet. It’s likely to be a nondescript, one-story building with no windows, since its functions are entirely automated. Also called local office, end office, local exchange, Class 5 office, switching office, and other more or less accurate things (but NOT “toll office”, which levels 1-4 of AT&T’s Class hierarchy are). Compare with POP.
The first three digits of a 7-digit US phone number identify the central office that serves that number. If it’s a call to the same exchange area, the responding CO switches the call to the destination CO. (Note that it’s not always that simple.)
Circular Error Probable. The radius of a circle within which 50% of rocket or missile launches (excluding failures) land. This number is used as a measure of accuracy.
Conférence Européenne des administrations des Postes et Télécommunications (used to be Comité Europeanne de Poste et Telegraphe). The organization responsible for the TDM data frame standards L1, L2, etc., which use symmetrical binary coding. The CEPT L1 (aka E1) is 256 bits/frame, 32 VGC time slots at 64 kb/s each (2.048 Mb/s total). Slot 1 is for synchronization and slot 16 typically for system signaling. The L2 is four L1s plus overhead, and the L3 is four L2s plus overhead.
Organisation (previously Conseil) Européenne pour la Recherche Nucléaire. The European particle research lab near Geneva where the WWW began. Its Large Hadron Collider (LHC) is the world’s most powerful particle accelerator. CERN is also the lead partner in developing DataGrid, an international project for exchanging and processing huge quantities of research data on the Internet.
Compact Flash. A small, removable, mass-storage card standard derived from the PC card, with a capacity up to 64 GB (2008). The card has an intelligent on-board controller, and can emulate an IDE or SATA hard drive without requiring the PC to have a special driver.
Chlorofluorocarbon. A non-flammable, non-toxic compound of carbon, fluorine, and chlorine. Such compounds, especially Freon™, were widely used as aerosols, solvents, and refrigerants in the 20th century, but were found to be destructive to stratospheric ozone. Production ceased in the mid-1990s as industries switched to HFCs.
Clip-bonded FlatPack. A surface-mount IC form factor created by Nexperia for circuits that need both high power and high density. Instead of wire traces, a copper clip inside the package connects its surface-mount contacts to interior circuit points. It reduces lead inductance and board footprint, and improves heat dispersion.
Common Gateway Interface. A standard process by which a Web server accepts user service requests and returns data. It employs scripts that parse user input strings and use them to generate HTML output. Malware creators are always hunting for vulnerabilities in the way widely used CGI scripts process this input.
The introduction of redundancy into data prior to transmission for the purpose of detecting or correcting errors at the receiver. Contrast source coding. See FEC, parity bit.
Challenge-Handshake Authentication Protocol. A security routine under PPP. The PPP server sends a challenge value to the remote unit attempting a connection. The remote unit returns a value calculated using a CHAP algorithm. Ideally, the challenge value is unique and not re-used.
The impedance Z0 that a cable would have if it were infinitely long – that is, if the surge impedance that a voltage source “sees” when it’s first connected across the signal and return terminals of the line were to be sustained indefinitely. This is the fundamental property of a transmission line. Z0 is a complex number, having a resistive and a reactive component. It’s a function of frequency, but is independent of line length. At very high frequencies, it asymptotically approaches a fixed (resistive) impedance, and this is the value used as the line’s characteristic impedance.
A perfect impedance match between a line and the terminating load (see Smith chart) would eliminate reflection of signal energy at the boundary, setting VSWR to unity. In practice, this isn’t achievable, but minimizing VSWR does minimize noise.
Impedance matching is why most coaxial cables are designed to have a high-frequency resistive component of either 50 or 75 Ω: the lowest transmission losses for air-dielectric coaxial cable occur at matched impedances around 70-80 Ω, while the best power handling occurs at 30 Ω. The 50 Ω standard is, therefore, a general-purpose compromise, while 75 Ω is preferred for large networks (e.g. cable TV) where minimizing transmission loss matters most. Other standards exist for specific applications, e.g. 300 Ω for folded dipole antennas; see RG##.
A value generated from digital data and stored or transmitted with it for error-checking purposes. If other users run the same algorithm on the data and get a different checksum, they know errors have been introduced. A checksum is smaller and faster than a hash, but not as robust, and is not intended to protect against deliberate tampering. It can’t flag multiple errors like a CRC, let alone find and fix them like the far more sophisticated forward error correction.
A parity bit is the simplest type of checksum. Another common approach divides the data into fixed-length data words and runs a serial XOR on them to produce a checksum word.
An IC chip containing some but not all of the components of a microprocessor or system on a chip. The use of chiplets enables manufacturers to use a smaller, faster process for parts of a design, while leaving harder-to-shrink sections on an older process. It also lets them improve a design without having to redo the whole thing. It’s one of the ways they’re trying to deal with the accelerating cost and difficulty of making ever-smaller circuits, while getting away from the need for larger die size.
High-bandwidth connections are used to link chiplets on a single package. The first industry standard for these connections, UCIe, was announced in 2022. Research continues on optical fiber interconnects.
A set of specialized PC motherboard IC chips connecting the processor to everything else, and providing support for basic motherboard features: memory, buses, keyboard, mouse, video, etc. The processor manufacturer commonly makes some but not all of them. There’s at least one chipset for each processor architecture. A motherboard’s capabilities are limited by the motherboard design as well as by the processor and chipset.
Intel’s Hub Architecture, standard in the 1990s and the 2000s, divides the PC chipset into a northbridge and a southbridge. The Platform Controller Hub (PCH) architecture introduced in 2008 puts most northbridge functions into the CPU, and the remainder plus all of the southbridge functions into a new PCH module. In some cases, the PCH is also integrated into the CPU die.
Continuous Integration/Continuous Delivery. The DevOps approach to software development: frequent changes to existing software (integration), which are immediately tested and pushed out as new releases (delivery) by automated processes.
CMOS-Integrated Nano-Photonics. The combination of optical components with CMOS on a single, monolithic IC chip, permitting IC designs that use the vastly greater data transfer speed of optical interconnections. IBM announced 90-nm process prototypes for commercial CINP chips in December 2012.
Cross-Interleave Reed-Solomon Code. The error correction scheme used in CDs. In order of priority, it applies Reed-Solomon error-correction coding (with interleaving), erasure correction, interpolation, and muting to compensate for bad read data. The latter two are for audio CDs only.
A 3-port RF device, usually ferrite-based, for matching and tuning antennas, amplifiers, and oscillators. RF energy input to a port will be output from one and only one other port (input to output, output to load, and load to input). A common application is the transmitter on input, the antenna on output, and the receiver on load. In this role, it’s also called a duplexer, since it allows transmission and reception to share an antenna. It isolates the transmit and receive signals from one another, and also isolates the transmitter from any reflection due to impedance mismatch with the antenna. If the load port is terminated, internally or externally, the device becomes a 2-port isolator, used for RF isolation between devices in a chain.
Cybersecurity Information Sharing Act. A 2015 law that permits government agencies, without obtaining a warrant or providing a specific reason, to receive personal data about US citizens from companies. It shields companies from liability for data-sharing.
Complex Instruction Set Computing. The name given to the original microprocessor design approach for Intel’s very successful x86 series: a large set of often-lengthy machine language instructions that support fairly direct mapping from high-level language instructions. This results in smaller application size, but more work for the processor – a logical tradeoff for early computers, with their small storage capacity and slow memory.
As memory and storage improved, other manufacturers in the late 1980s and 1990s took the rival RISC approach, which promised better performance by lightening the load on the CPU. Each camp has borrowed useful features from the other, and the constraints on computer performance have continued to change, so the CISC/RISC distinction has faded.
Contributor License Agreement. An agreement by contributors to a software project to grant specific rights to other developers who use their work. See GPL.
Command Line Interface. A program interface that requires users to type in commands, as opposed to a GUI. Early PC operating systems (see DOS) used a CLI, but could run GUI-based software.
The model used for most computer networks, including the Internet. It categorizes network nodes as either server nodes that have data, or client nodes that access the data, and uses a centralized database to regulate these roles. It offers better security and scalability than peer-to-peer networking, but is more expensive and less flexible.
Performing computing and data-storage tasks with Internet resources instead of with local software and hardware. It’s a return to the 1960s-1970s concept of time-sharing, when powerful computers were too expensive for most users to own. Maintaining, protecting, and upgrading modern computing resources and efficiently sharing them among multiple users is a complex task. Outsourcing it to specialists rather than doing it in-house is often more cost-effective, and arguably safer.
Common models of cloud computing:
Software as a service (SaaS) – The cloud provides and runs the applications that clients use.
Platform as a service (PaaS) – The cloud provides hardware, real or virtual, for applications and data that are provided and controlled by clients.
Infrastructure as a service (IaaS) – The cloud provides virtual hardware, storage, and networking, and the client provides the operating system and everything in it.
Edge computing is a subtype of cloud computing that reduces latency and security risks by using computing resources that are as close as possible to the customer’s physical location. So-called serverless computing does still use servers, but abstracts the process of allocating them to customers based on real-time demand, similar to how packet-switching makes more efficient use of transmission bandwidth as compared to circuit-switching.
The biggest cloud services as of 2015 are Amazon.com’s AWS and Microsoft’s Azure.
Common Language Runtime. Generically, a computer runtime environment that enables platform-independent program code to run on specific hardware. Microsoft gave the name CLR to the environment for its .NET development toolkit, so most uses of the term mean that specific application.
Common Mezzanine Card. The IEEE P1386 standard, which specifies a mechanical form factor for mezzanine cards (a.k.a. daughtercards). The PMC standard is based on CMC.
Coronal Mass Ejection. A huge bubble of ionized, super-heated gas (plasma) ejected by the Sun. The frequency of CMEs varies with the approximately 11-year solar cycle, from about once a week at solar minimum to an average 2-3 times a day at solar maximum. On the rare occasions when a CME is aimed directly at Earth, the radiation burst takes about 9 minutes to arrive, and can damage satellite electronics. The energetic ions from such a CME take a day or two to arrive, but are more destructive. They cause high current flows in the ionosphere, inducing ground currents (GIC = geomagnetically induced current) that damage unprotected power lines, transformers, and grounded electronics. Most electrical infrastructure worldwide is unprotected.
Capability Maturity Model Integration. A process management and improvement model that many software development teams follow. It defines five levels of increasing process maturity: Initial (meaning ad-hoc development), Managed, Defined, Quantitatively Managed, and Optimizing. Each level except the first has process areas (PAs) defining the activities that it requires. Level 4 emphasizes a quantitative (measurable) approach to both the process and the product, and Level 5 aims to measure improvement in both, with quantitative goals and performance tracking.
As a counterpoint, there’s a Capability Immaturity Model (CIMM) that defines levels from 0 to -3: Negligent, Obstructive, Contemptuous, and Undermining.
Complementary Metal-Oxide-Semiconductor. A very common type of IC based on complementary pairs of n-channel and p-channel MOSFETs, which makes it more power-efficient than NMOS or PMOS alone. The term can mean 1) the chip technology itself, 2) the digital logic standard that such chips use (see logic family), 3) a CMOS-based, battery-backed NVRAM chip in older PCs that stores configuration data about the system and supplies it to the BIOS at boot-up, or 4) a CMOS-based photodetector technology used in most digital cameras (compare CCD). Newer PCs use flash memory or EEPROM for both the configuration data and the BIOS itself.
Communications Plenum. A National Fire Protection Association standard for the jacket of plenum cable, which must be fire-resistant like CMR and also must not produce toxic smoke if burned.
Communications Riser. A National Fire Protection Association standard for the fire-resistant jacket of riser cable, meant for use in building walls and vertical risers.
Content Management System. Software that presents content (whatever that might be) in some desired format, and manages changes to both content and presentation. This describes stand-alone apps such as Web development platforms, as well as multi-user systems for coordinating work.
Obsolete connector standard from Centronics (hence the CN). The number is the number of pins/sockets in the connector. CN36 was commonly used for printers before USB and FireWire, and CN50 was for SCSI1. A variant type, HPCN##, includes HP for Hewlett-Packard. HPCN36 is a IEEE-1284 type C parallel port connector once used for laser printers.
A coaxial cable is an unbalanced transmission medium, with the signal conductor sheathed in a dielectric and an outer sleeve of foil or wire braid. This second conductor is the signal return, not a shield – although it’s often grounded at both ends, and sometimes to intermediate points such as bulkheads.
Coaxial cable
It’s better to isolate it from ground at intermediate points to avoid ground loops and common mode noise. It’s best to ground it only at the source. Grounded coaxial cable can safely carry most signals from 20 kHz to 5 GHz, although better cables are strongly recommended for higher frequencies. Impedance matching is important for relatively narrow frequency ranges. The most common cable connector type for coaxial cable is the BNC, but there are several others in wide use.
Coder/decoder. A device or application that encodes signals from some source into digital data, and decodes such data back into the original source form. If the source is analog, the codec does analog-to-digital and digital-to-analog (A/D and D/A) conversion too. Often, the coder provides data compression to reduce the transmission rate or file size, and the decoder then handles decompression (see definition (2) below). A/D conversion and compression can be combined into one process.
Some popular types of video codecs, i.e., standards for recording, compressing, transmitting, decompressing, and playing video with sound:
AV1 – (2018) Aoemedia Video 1. A royalty-free, open-software format with compression rates as much as 50% higher than H.264 and roughly on par with H.265.
H.263, H.264, H.265 – Closed, royalty-encumbered formats released by the ITU-T and partners in 1996, 2003, and not-quite-2018. As of 2018, Blu-Ray DVD players use H.264, and so do most Internet videos.
Compressor/decompressor. Software that compresses digital data so that it requires less disk space or bandwidth, and decompresses it for use. The term “codec” isn’t used for this function anymore, since it invites confusion with (1).
Compression is characterized as either lossless, meaning the compressed data can be restored to exactly its original form, or lossy, meaning it can’t. General-purpose file-compression utilities such as PKZIP must use lossless compression. On the other hand, data representing an analog source (image, sound, video, etc.) can suffer some loss of fidelity from lossy compression and still be acceptable. This is typically done not by a stand-alone app but as a function of whatever captures the source data.
Common Object File Format. A format for binary code files – program executables, object files, and DLLs or other libraries. The files can have the .cof or .coff filename extension, but often have a different extension to declare their content, such as .bin for binary or .o for object. Microsoft uses a variation of COFF rather than the newer ELF for its Windows .dll and .exe files.
Demodulating a communications signal carrier using reference signals that exploit detection of the carrier phase to achieve phase lock. This requires a matched filter. It’s more complex than non-coherent detection, but has better performance. Coherent modulation controls the instantaneous phase of the signal.
Component Object Model. This confusing Microsoft “standard” began in 1993 as the communication model for OLE 2. It allows applications to share data and command objects, but only in Windows. It uses a compiled IDL (interface definition language) to define COM object interfaces, which are a set of function calls bound by a GUID (globally unique identifier). In 1996, Microsoft folded COM into ActiveX. In 1997, they named the whole mess COM again. With the introduction of Windows 2000 Professional, COM borrowed features from DCOM and became COM+, before being phased out for .NET. See OLE, ActiveX, DCOM.
Communications. A COM port on a computer is an RS-232c serial data input/output port. Early PCs could have four COM ports internally, but only two interrupt requests (IRQ 3 and IRQ 4) to support them. As a result, they had just one or two physical COM ports, typically DE-9P connectors. Modern PCs are capable of having more, but often don’t have any – USB and wireless connections are more popular.
Commodore International was an innovative early player in the personal-computer market. Mismanaged, and eventually eclipsed by Apple and x86 vendors, they went bankrupt in 1994. The devotion of brand loyalists still (2010) drives a steady trickle of upgrades and new products, some supporting Linux, Windows, and Mac OS X. A couple of their classic products:
Commodore 64 – (1982) The best-selling single model of personal computer ever. 8-bit, 985 or 1023 kHz processor, 64kB of RAM, 16-color display.
Amiga 1000 – (1985) The first multimedia personal computer, using a hybrid 16/32-bit Motorola 68k processor and the multi-tasking Amiga OS. Its superior graphics performance depended on a dedicated graphics coprocessor, and on synchronizing its system clock to an NTSC video signal.
A type of out-of-band signaling that uses a separate control network for network services and operations, as well as for management of individual connections.
Interference that derives from a single noise source entering a system at multiple points, with a different potential at each point. Having just one system ground can minimize this.
Also called cPCI. A 1995 local bus standard supported by PICMG, based on Eurocard packaging and the PCI electrical specification. It supports 8 slots without bridging instead of 4. The space-saving design is meant for industrial applications, and is common in rack-mount and high-end systems. The modules are designed for front loading and removal from a card cage. They’re held in position (usually vertical) by the connector, card guides, and a screw-down face plate.
cPCI female (board side) J1 connector
CompactPCI modules are made in 3U (100×160mm) and 6U (160×233mm) sizes, and use the IEC-1076 international standard 2mm pin/connector interface. The 3U module uses 220 pins/sockets in 5 rows, as shown above (the J1 connector), and supports 32- or 64-bit PCI. Twenty of the pins are reserved for future use. Boards that use only 32-bit PCI can get by with just 110 pins/sockets. The 6U boards can have up to 440 pins (two J1 connectors), but the additional pins have been deliberately left undefined. See PCI.
Compressing & Expanding. In quantizing digital samples of some source to produce a signal, this means assigning smaller quantization steps to the more important ranges of the source. For telephone voice signals, North America & Japan use µ-law 8-bit logarithmic companding with a µ factor of 255 (linear amplification is at µ = 0). It’s optimized for good signal-to-quantization-ratio over a wide dynamic range. Europe uses A-law 7-bit companding with an A factor of 87.6 (linear amplification is at A = 1). It’s optimized for a more constant signal-to-quantization-noise ratio with less dynamic range.
From “Communications Satellite Corporation”. Established by the US government in 1962, made public in 1963, and now part of Lockheed Martin. Its purpose was to develop commercial satellite communications systems. It led the creation of Intelsat.
The use of two kinds of channel coding to improve BER performance. In satellite comms, the convention is to first apply a Reed-Solomon code, and then a convolutional code on top of it. Of course, the second code applied must be the first decoded at the receiver. Common practice is to call the first code applied the outer code, and the second the inner code, reflecting a system-flow rather than a signal-processing perspective.
Signals that avoid amplitude level shifts. This enables them to use the more efficient saturated and non-linear power amplifiers without distortion, which is valuable for satellite comms. The cost is an increase in the bandwidth of the signal’s main lobe. M-ary PSK signals are technically constant-envelope, but the filtering required to remove their spectral sidelobes makes them non-constant, and a non-linear amplifier that restores constancy also restores the sidelobes. Some PSK variants such as OQPSK are constant-envelope, as are FSK, MSK, and their variants. QAM, obviously, is not. See CPM.
A common way of viewing digital signal quality in real time, also called an I/Q constellation, polar plot, or Argand diagram. It’s a special-case Lissajous pattern in which the oscilloscope inputs are the in-phase (I) and quadrature (Q) rails of a phase-modulated signal. With I as the X-input and Q as the Y-input, a digital o-scope synchronized to the symbol rate displays the I amplitude on the abscissa (horizontal) and the Q amplitude on the ordinate, so each received symbol appears as a phase-angle/amplitude point on a polar graph, as shown below. See eye diagram.
In software terms, a program that provides a consistent runtime environment for other programs, permitting them to run on different platforms. It starts much faster and consumes far fewer system resources than a virtual machine, but is considered less secure.
The Open Containers Initiative is the Linux community’s project to standardize containers. As of 2023, Docker is the most popular open-source container software, and Kubernetes, abbreviated K8s, is the most popular open-source software for deploying and managing containers on cloud servers.
A Web site that uses poor-quality content crafted to maximize the site’s selection by search engines. These sites are usually trying to make money by serving ads.
A device that changes the digital and/or analog properties of a signal from some source or device so that another device can use it. “Signal” in this case can mean the voltage, current, and frequency of electric power. Converters for data signals commonly work in both directions, whereas most power converters are one-way. Not the same thing as an adapter.
From magic cookie, a pre-Web term for the concept. A small (up to 4 kB) text file that a Web browser stores on the user’s computer to hold information about the Web site that the user is accessing. The browser sends the cookie to the Web site with each HTTP request, and then displays the site page with whatever state settings the cookie holds. These typically include such things as scrolling position, screen resolution, and selections made by the user. It also contains user authentication if the site uses that.
Third-party cookies are pushed on visitors to a Web site by some party, commonly an advertiser, that has an agreement with the Web site’s operators permitting them to do this. They’re typically for user tracking and serving targeted ads.
Cookies have obvious privacy/security issues. Many users now (2014) block their browsers from submitting them, although this affects the usability of some Web sites. Companies that are determined to track users have other options, such as browser fingerprinting: directing the browser to draw a hidden image with JavaScript, something each PC will do differently due to its particular combination of browser, OS, font, display settings, clock time, etc. Certain ISPs support tracking by injecting user-specific serial numbers into Web requests at the network level (layer 3 of the OSI model); the only defense is an encrypted connection, preferably through a VPN or a darknet account.
Named for American physicist Leon Cooper. The basis for superconductivity: two electrons weakly bound to one another in such a way that they move without resistance through a conducting medium. The weak nature of the bond means that too much energy or the wrong sort of medium will break the pairs, which is why superconductivity has been observed only at low temperatures and in certain materials. The BCS theory of superconductivity (B for John Bardeen, C for Cooper, S for Robert Schrieffer) holds that Cooper pairs form through interaction with the crystal lattice of the medium. The dominant view is that the pairs align with their spins in opposition, and that this anti-ferromagnetic (magnetically neutral) state, by preventing magnetic interaction with the medium, is what permits superconductivity.
Common Operating Picture. A system that presents the same real-time view (map or otherwise) to multiple users, especially users in different physical locations.
Common Object Request Broker Architecture. A platform-independent distributed component model. That means it allows applications to present components, such as commands and data objects, that applications on other computers can use. It was designed from the start to operate across the Internet. Compare OLE, COM, ActiveX, DCOM.
Coordinate Rotation Digital Computer. A computation-saving 1959 algorithm for trigonometric and transcendental functions using only add and shift operations. HP put it to work in digital calculators.
Named for French scientist Charles-Augustin de Coulomb (1736-1806). The force F between two point charges Q1 and Q2 is proportional to the product of the charges and inversely proportional to the square of the distance r between them: F = k [Q1 Q2] / r2. The term k is a constant of proportionality defined by k = 1/4πε, where ε is the electrical permittivity of the medium. If the two charges are of the same sign, the force is outward (repulsive).
Resonant circuits, such as filters, are often used in series to improve attenuation at target frequencies. They must be coupled together. The coupling mechanism varies with the application. The four most common types are series capacitive, series inductive, transformer (mutual inductive), and active (transistor) coupling.
Continuous Phase Modulation. A non-linear, constant-envelope digital modulation technique common in wireless modems. It can be viewed as either PSK or FSK, since it carries information only in its phase or frequency. It uses Gaussian, raised cosine, or partial-response filtering of data pulses prior to modulation to prevent phase discontinuities between signal states. This reduces sidelobe spectral energy and out-of-band interference, and improves co-channel performance. MSK and GMSK are common types of CPM. CPM has memory, so its demodulation requires maximum-likelihood sequence estimation (see Viterbi algorithm).
Cyclic Redundancy Check. An error-detection method for blocks of binary data, more robust than use of a checksum. CRC algorithms are denoted CRC-n, where n is the bit length of the checksum and also the order of the generating polynomial – but remember, an n-order polynomial has bit length [n+1].
The standard method of implementing a CRC is to append n zeros to an m-length data word G1(x), and then, starting at the highest “1” bit, XOR this [m+n]-length block with the generating polynomial P(x) of order n. Repeat this XOR process until the remainder is of order n-1 or less, at which point the XOR can’t continue. The n least significant bits of the remainder are the checksum for G1(x). Append the checksum to G1(x) to produce the CRC block F1(x) for transmission or storage. To check for errors in each F(x), XOR it cyclically with P(x), as above; if the data has no errors, there should be no remainder.
There are many standard implementations of a CRC. The CCITT’s CRC-16 uses P(x) = [x16 + x12 + x5 + 1], which equates to the 17-bit sequence [10001000000100001]. The IEEE’s CRC-32 uses P(x) = [x32 + x26 + x23 + x22 + x16 + x12 + x11 + x10 + x8 + x7 + x5 + x4 + x2 + x + 1], or [100000100110000010001110110110111]. See GP, LFSR.
Color Rendering Index. A measure, up to 100, of how well a light source’s color matches natural light. LED lights must have a CRI of at least 80 to receive DOE Energy Star approval. Incandescent and halogen lights have a CRI of 100, as would radiation from a blackbody at a temperature of 6500 K.
Clustered Regularly Inter-Spaced Palindromic Repeats. Way outside the scope of this site, but it’s important: a gene-editing molecular toolkit of enzymes, target-mirroring ribo-nucleic acid (RNA) sequences, and DNA-cutting proteins such as Cas9. CRISPR allows its users to replace or simply remove specific gene sequences within a cell.
Carriage Return/Line Feed. In computer representations of text, a character – or two characters – that indicate display should return to the beginning of the line and also go down one line, thereby starting a new line. That’s why it’s also called a newline. The terms originate in the mechanical details of typewriters and teletype machines, which, for various reasons, could use CR and LF together or separately. This has led to different operating systems, and sometimes different programs for the same operating system, implementing newlines in different ways. MS Windows (and MS-DOS) programs normally use CR/LF. Programs for Apple OS X, Unix, and Unix descendants normally use just LF. Some apps react badly to one or the other, and some apps can handle either style. Few if any modern programs use just CR.
In Unicode, CR and LF have the decimal values 0013 and 0010 (hex 0x000D and 0x000A) respectively. They’re also commonly denoted as \r and \n, the escape sequences used for them in C and other programming languages.
A CRLF injection attack inserts malicious code, preceded by CR/LF characters, into a server’s input field. If the server isn’t screening for this, it will treat what comes after the CR/LF as a new and valid instruction.
Also called a null-modem cable. A cable that permits two DTEs (computers) with Ethernet or serial RS-232 ports to communicate without going through a DCE (modem, hub, router, switch). Also sometimes used to directly link two DCEs. The table below shows the correct pin-to-pin wiring for serial D-sub cables and Ethernet Cat# 8P8C cables, with * indicating a pin that is sometimes left unconnected. Compare patch cable.
A mathematical operation used in analyzing the behavior of EM fields, among other things. Given two vectors, their cross product is a vector orthogonal (using right-hand rule) to both, with magnitude equal to the product of their magnitudes and the sine of the angle θ between them. For example, a × b = –b × a = n ab sin(θ), where n is the unit vector orthogonal to vectors a and b. The × operator is easily confused with multiplication, but remember, you can’t just multiply two vectors. Compare with dot product.
An intentional short-circuit that shunts current to ground to protect the system from an overvoltage condition. The name is a comment on the crudeness of the approach, equivalent to laying a steel bar across the output terminals.
Cathode Ray Tube. A type of thermionic valve with magnetic or electrostatic beam-steering, and a fluorescent screen that displays the electron beam as an image. It was the basis for not only all TVs and PC monitors but most forms of video displays, until the advent of the flat panel display (FPD).
Crypto, for short. A digital Internet currency with no controlling authority. Bitcoin was the first, in 2009, but now there are thousands of others. Most, but not all, are based on blockchain algorithms. Many are scams. See also NFT.
Carrier Sense Multiple Access/Collision Detection. A bus network technology using the MAC protocol. It’s the basis for the Ethernet (IEEE-802.3) standard. Each station on the network listens for traffic while transmitting. If another station transmits simultaneously, this is a collision, and the station will pause for a random period of time before attempting transmission again.
Chip Scale Packaging. A surface-mount IC in a package that, per the JEDEC definition, has board surface no more than 20% larger than the bare chip. See also WLP.
Concentrated Solar Power. An energy technology that uses arrays of reflectors to concentrate sunlight on a heating element, which uses the heat to drive a turbine. Some designs use oil or molten salt to absorb the heat and exchange it with water; others use only water, a simpler but less flexible approach.
Cross-Site Request Forgery. Pronounced sea-surf. A type of malware attack: a malicious site spams script or HTML commands for common Web applications that site visitors might be running. If a visitor is running a vulnerable application, the hostile site can gain access with the user’s own privilege level.
Cascading Style Sheet. A W3C standard for applying formatting to XML-based (HTML, XHTML, etc.) Web documents by putting it in a separate .css file. This eliminates most of the clutter and repetitive labor of HTML format tags, reduces file size and load time, and greatly simplifies changing Web site appearance. “Cascading” refers to how browsers can apply styles from multiple CSS sources, with the most recently applied source taking precedence over previously applied ones if there’s a conflict.
The CSS file contains sets of formatting rules for browsers to apply in displaying HTML elements. A set of rules can be defined to apply to all instances of a particular type of element, or of more than one type of element. A rule set can also be defined as a class or an ID, in which case the HTML code for an element must specifically invoke that class or ID to receive the formatting.
A class is defined in CSS preceded by a period, e.g. .class_name, and invoked in HTML by a construction such as <p class="class_name">. The CSS definition can attach the class to a specific HTML element, in which case only elements of that type can use it; or it can be a standalone definition that any element can invoke.
An ID works much like a class, except that its definition uses the hash character # instead of a period, and its invocation uses id= instead of class=. The functional difference is that each ID is meant to be used only once on any given Web page. It’s often used for search or linking purposes, with no associated CSS formatting.
Channel Service Unit. A device on digital links that transfers data in the 56 kb/s to 1.5 Mb/s range, and performs line-conditioning and equalization functions. It connects digital lines from a carrier’s central office to devices such as routers and channel banks. Often combined with the DSU.
Control. Many keyboard shortcuts require holding down the Ctrl key while pressing another key. MS Windows shortcuts include select all (Ctrl-a), copy (Ctrl-c), cut (Ctrl-x), paste (Ctrl-v), and undo last action (Ctrl-z). Individual applications have additional shortcuts, such as find (Ctrl-f), find & replace (Ctrl-h), new document (Ctrl-n), open document (Ctrl-o), save (Ctrl-s), and print (Ctrl-p).
The copy and paste shortcuts, Ctrl-c and Ctrl-v, are sacred to the Church of Kopimism, a file-sharing religion started in Sweden. Really.
Computerized Tomography scan. Also Computerized Axial Tomography (CAT) scan, which is why CT is pronounced “cat” rather than “see-tee”. A medical technology for using X-rays from multiple angles, and an array of detectors instead of conventional film, to develop 3D images of subjects. Compare MRI, PET.
Consumer Ultra-Low Voltage. A type of very thin, low-power laptop that Intel introduced in 2008-2009, with TDP < 10W. It was later branded an ultrabook.
Chemical Vapor Deposition. A laboratory and industrial technique for applying a thin coating of a material. It relies on vacuum to create uniform distribution. Details vary with the material and the substrate, but the material being deposited is typically vaporized by heat or a chemical reaction.
Common Vulnerabilities and Exposures. Software industry term for the security vulnerabilities of a system or a software application. NIST maintains a record of known CVEs called the National Vulnerability Database (NVD) at https://nvd.nist.gov/. Each vulnerability has an ID with the format CVE-yyyy-nnnnn, where yyyy is the year it was discovered and nnnnn is an identifying number from the CVE Numbering Authority (CNA) that made the discovery.
Cockpit Voice/Flight Data Recorder. Most (all, in the US) commercial aircraft carry systems that record data during flight, and that are intended to survive a crash to aid investigators. All such systems are popularly called black boxes, although they’re most often orange. To increase their chances of survival, they’re usually in the tail section. They have two functions:
CVR – Cockpit Voice Recorder. Records the last 30 to 120 minutes of voice data from up to four cockpit sources.
FDR – Flight Data Recorder. A flight data acquisition unit (FDAU) collects data from on-board sensors. The FDR stores up to 25 hours of this data on magnetic tape or solid-state memory in its crash-survivable memory unit (CSMU).
Some systems house the two separately. Others combine them into a single physical unit, the CVFDR, or have redundant units. Each unit has an underwater locator beacon (ULB) on the outside; if submerged, it begins transmitting 37.5 kHz sonic pings at the rate of one per second, and continues for a few weeks until its battery fails.
Concurrent Versions System. Open-source version-control software dating to 1986. A more capable implementation of CVS called CVSNT was released under GPL in 1998. WinCVS is a MS Windows freeware GUI application that installs and manages CVSNT. Compare Subversion.
Card Verification Value. Also card security code (CSC). A 3- or 4-digit number on the back of a credit card or debit card, to provide additional validation for online, telephone, or mail transactions where the owner can’t present the physical card.
An important sub-class of linear block code with the feature that, given a valid block codeword, any wrap-around shift of that codeword is also a valid codeword.
A numeric measure of the complexity of a piece of software, based on the number of distinct execution paths. Software with high CC is more fragile, more likely to contain flaws, and harder to understand, change, or test.
Initially an x86 math coprocessor company founded in 1988 by ex-TI employees, Cyrix moved into microprocessor ICs. It was acquired by National Semiconductor in 1997, run into the ground, and then bought by VIA Technologies (Taiwan) in 1999. Some of its products:
5x86 (M1sc) – 1996 clock-tripled (100 or 120 MHz), 4th or 5th generation CPU, normally for 33 or 40 MHz motherboards.
6x86 (M1) – 1997 6th-generation Pentium rival, much cheaper and slightly faster. 100 to 150 MHz clock speed; 64-bit I/O bus at half core speed, from 50 to 75 MHz (contemporary Pentiums topped out at 66 MHz), no MMX support. The need for a 75 MHz motherboard chipset limited its market.
6x86MX (M2 or MII) – A faster, MMX-capable version of the 6x86. Has 0.35 µm process, unified 64 kB L1 cache, clock speeds to 350 MHz, system bus speeds 66 to 100 MHz.
Cyrix III (M3 or MIII) – The first (2000) of this processor series used a Cyrix core, but later models are based on designs from Centaur, also bought by VIA from IDT. They use Socket 370 like the Pentium III, and have clock speeds 433-533 MHz and system bus speeds 66, 100, or 133 MHz.
