Engineer Information Communications common mistakes
Engineer Information Communications · Understand the traps and how to avoid them.
Computing a link's Shannon capacity or a channel's noise power with the wrong unit prefix, e.g. plugging a bandwidth given in MHz directly into C = B·log2(1+S/N) without converting to Hz, or treating a dB value as a linear ratio.
Candidates memorize the formulas in symbolic form and then rush the substitution, so they mix Hz/MHz, dBm/mW, and dB/linear when a question deliberately supplies values in mixed units to test dimensional awareness.
What to do instead
Before substituting, write every quantity with its SI base unit and prefix exponent (e.g. 20 MHz = 20×10^6 Hz; 30 dBm = 10^3 mW). Convert dB to linear with 10^(dB/10) for power and 10^(dB/20) for voltage/amplitude. Practise a fixed routine: list givens, normalize units, substitute, sanity-check magnitude (a 20 MHz channel cannot exceed ~20 Mbit/s per Hz of spectral efficiency).
Miscounting the number of subnets or usable hosts after subnetting an IPv4 prefix, forgetting that the all-zeros and all-ones host portions are reserved and that /31 and /32 are special cases.
The arithmetic of 2^n is easy, so candidates skip the last step and report the raw power of two instead of subtracting two, or they apply the old classful reservation rules to modern CIDR point-to-point links.
What to do instead
Use the standard formula: usable hosts = 2^(32−prefix) − 2 for prefixes /1 to /30; a /31 gives 2 usable addresses on point-to-point links (RFC 3021) and a /32 is a single host route. For subnet counts, remember that borrowing n bits yields 2^n subnets under CIDR, not 2^n − 2. Work two or three examples by hand each study session until the subtraction becomes automatic.
Confusing which OSI/TCP-IP layer a protocol or device operates at, e.g. calling a switch a Layer 3 device, placing TLS at the transport layer, or saying ARP is a Layer 3 protocol.
Real-world products blur layers (multilayer switches, TLS libraries inside applications), so candidates rely on product marketing rather than the formal layer definitions the exam uses.
What to do instead
Build a one-page layer map: Ethernet/802.11 = L1–L2, IP/ICMP/ARP = L3 (ARP resolves L3 to L2 but is commonly classed as L2.5/L3-adjacent; state the convention your syllabus uses), TCP/UDP = L4, TLS = between L4 and L7 (session/presentation in OSI terms), HTTP/DNS/SMTP = L7. For each device (hub, switch, router, firewall) note the highest header it inspects, which determines its layer.
Mixing up free-space path loss, received power, and SNR when solving a link-budget question, e.g. adding antenna gains but subtracting them, or omitting cable/feeder losses and fade margin.
The Friis equation and the dB form of the link budget look similar, and candidates memorize one version without tracking the sign convention of gains versus losses.
What to do instead
Write the budget as a single ledger: Pr(dBm) = Pt(dBm) + Gt(dBi) + Gr(dBi) − Lpath(dB) − Lfeeder(dB) − Lmisc(dB). Compute FSPL as 20log10(d) + 20log10(f) + 32.44 with d in km and f in MHz (or 92.45 with GHz). Keep gains positive and losses negative, and verify that received power at a realistic distance lands in the expected −60 to −110 dBm range.
Assuming a higher-order modulation or wider channel always improves throughput, ignoring the SNR/MCS threshold and the Shannon limit; or treating MIMO spatial streams as free capacity regardless of channel rank.
Marketing material equates bigger QAM numbers with better performance, so candidates pick 256-QAM in a low-SNR scenario where the correct answer is a robust lower-order scheme.
What to do instead
Tie each modulation/coding scheme to its required SNR (e.g. QPSK tolerates much lower SNR than 64-QAM) and remember the exam's rule: choose the highest MCS whose threshold the link meets with margin. For MIMO, state that spatial multiplexing requires sufficient channel rank and SNR; otherwise diversity or beamforming is the better answer. Recompute capacity with the given SNR to confirm the choice.
Confusing attenuation and dispersion (or chromatic vs. polarization-mode dispersion) when explaining why an optical link fails, or quoting a reach figure that ignores the dispersion penalty and OSNR budget.
Both impairments reduce signal quality, and candidates memorize 'dispersion limits distance' without distinguishing the mechanisms, so they misattribute a power-budget failure to dispersion.
What to do instead
Separate the two: attenuation is a power loss measured in dB/km and is fixed by amplifiers and receiver sensitivity; dispersion is pulse spreading that causes intersymbol interference and is fixed by compensation or coherent DSP. For a reach question, first check the power budget (launch power − fiber loss − connector/splice loss − margin vs. sensitivity), then check dispersion/OSNR limits, and state which one binds first.
Mixing up symmetric and asymmetric cryptography in use cases, e.g. saying RSA encrypts bulk data, that a digital signature provides confidentiality, or that hashing alone provides authentication.
The primitives are taught together and candidates remember 'public key = secure' without mapping each primitive to the security property (confidentiality, integrity, authenticity, non-repudiation) it actually provides.
What to do instead
Memorize the mapping: symmetric (AES) for bulk confidentiality and speed; asymmetric (RSA/ECC) for key exchange and signatures; hash (SHA-2/3) for integrity only; HMAC for integrity plus authenticity with a shared key; digital signature for authenticity and non-repudiation. When a question asks 'which property', name the property first, then the primitive. Note that encryption alone does not prevent tampering without an integrity check (AEAD).
Treating TLS certificate validation as automatic or assuming a valid certificate proves the server is trustworthy, and misstating what a CA signature actually guarantees.
Candidates conflate 'the padlock is present' with 'the site is safe', and exam items probe the distinction between identity binding, chain of trust, and server intent.
What to do instead
State the chain explicitly: the server presents a certificate binding its public key to a name; the client verifies the signature chain to a trusted root, checks validity dates, name match, and revocation. A valid certificate proves the key belongs to the named entity, not that the entity is honest. Also recall that TLS provides confidentiality and integrity in transit only, not at rest.
Predicting program output without accounting for pass-by-value vs. pass-by-reference, short-circuit evaluation, integer division/overflow, or operator precedence in the language the exam uses.
Candidates read code as pseudocode and apply their mental model from another language, so they miss that, for example, a function receives a copy of a reference or that && skips the right operand.
What to do instead
For every code question, annotate the language and its semantics: does the call pass a copy (Java/C pass references by value; C++ can pass by reference), what is the integer width and signedness, and where do side effects occur? Trace with a table of variable values line by line, and explicitly evaluate short-circuit operands left to right. Confirm the language standard the syllabus specifies before assuming behavior.
Answering regulatory items from general memory instead of the specific framework, e.g. confusing ITU-T and ITU-R scope, mixing spectrum licensing classes, or misstating which body allocates national numbering and frequencies.
Telecom law is layered (international treaties, regional bodies, national regulators), and candidates collapse the layers into one 'the government regulates it' answer that fails on the specific body or instrument.
What to do instead
Build a hierarchy table: ITU-T = standards for telecommunication systems; ITU-R = radio spectrum and satellite orbit matters; national regulator = licensing, numbering, type approval, and enforcement within national law. For spectrum questions, distinguish licensed, license-exempt, and shared-access regimes and name the instrument (allocation table, license conditions). Verify any specific article or clause number against the current text before relying on it.