Communication Systems Interview Questions

Modulation is the process of varying a carrier signal's properties (amplitude, frequency, or phase) according to the information signal. It is necessary because: baseband signals cannot be transmitted efficiently over long distances, antenna size must be comparable to wavelength (impractical for audio frequencies), it enables frequency division multiplexing for multiple channels, and it shifts signals to frequency bands with better propagation characteristics.

In AM (Amplitude Modulation), the carrier amplitude varies with the message signal while frequency remains constant. In FM (Frequency Modulation), the carrier frequency varies while amplitude stays constant. AM is simpler and requires less bandwidth but is susceptible to noise and interference. FM provides better noise immunity (amplitude variations can be limited), better audio quality, but requires wider bandwidth. AM is used in long-wave broadcasting; FM is used for high-fidelity audio and mobile communications.

Bandwidth is the range of frequencies a communication channel can transmit, measured in Hertz (Hz). It determines the maximum data rate a channel can support (Shannon's theorem: C = B*log2(1+SNR)). Greater bandwidth allows higher data rates and better signal fidelity. For analog signals, bandwidth is the range from lowest to highest frequency component. For digital signals, bandwidth relates to symbol rate and modulation scheme. Bandwidth is a precious resource, allocated by regulatory bodies.

The Nyquist sampling theorem states that to accurately reconstruct an analog signal from its samples, the sampling frequency must be at least twice the highest frequency component in the signal (fs >= 2*fmax). This minimum rate is called the Nyquist rate. If sampling is below this rate, aliasing occurs where high-frequency components appear as lower frequencies, causing distortion that cannot be removed. Anti-aliasing filters are used before sampling to remove frequencies above fs/2.

Simplex communication is unidirectional, data flows in one direction only (e.g., broadcast radio, TV). Half-duplex allows bidirectional communication but not simultaneously; one party transmits while the other receives (e.g., walkie-talkies, some Wi-Fi). Full-duplex enables simultaneous two-way communication using separate channels or frequencies (e.g., telephone, cellular networks). Full-duplex requires more resources but provides better user experience and higher throughput.

Signal-to-Noise Ratio (SNR) is the ratio of signal power to noise power, typically expressed in decibels: SNR(dB) = 10*log10(Ps/Pn). Higher SNR indicates cleaner signal with less noise interference. SNR directly affects communication quality and achievable data rates (Shannon capacity). It determines bit error rate in digital systems. SNR is affected by transmit power, distance, interference, and receiver noise figure. System design aims to maintain adequate SNR throughout the communication link.

ASK (Amplitude Shift Keying) varies carrier amplitude to represent digital data; simple but noise-susceptible. FSK (Frequency Shift Keying) uses different frequencies for different symbols; good noise immunity but requires more bandwidth. PSK (Phase Shift Keying) varies carrier phase; efficient bandwidth usage and good noise performance. BPSK uses two phases (180 degrees apart), QPSK uses four phases encoding 2 bits per symbol. PSK and its variants are widely used in modern digital communications.

Baseband transmission sends the original signal directly over the medium without modulation, using the full channel bandwidth starting from DC (0 Hz). It is suitable for short distances like LANs. Passband (broadband) transmission modulates the signal onto a carrier frequency, shifting it to a specific frequency band. This enables long-distance transmission, frequency division multiplexing, and matching signal to medium characteristics. Examples: Ethernet uses baseband; cellular and satellite use passband.

Multiplexing combines multiple signals for transmission over a single channel. FDM (Frequency Division Multiplexing) assigns different frequency bands to each signal, used in radio/TV broadcasting. TDM (Time Division Multiplexing) assigns different time slots to each signal, used in digital telephony. WDM (Wavelength Division Multiplexing) uses different light wavelengths in fiber optics. CDM (Code Division Multiplexing) assigns different codes to signals, enabling simultaneous transmission at same time and frequency (CDMA).

A superheterodyne receiver converts the incoming RF signal to a fixed intermediate frequency (IF) before demodulation. It consists of: RF amplifier, local oscillator, mixer (produces IF = RF - LO), IF amplifier and filter, and demodulator. Advantages: Selectivity and gain at fixed IF are easier to optimize, tuning requires only changing LO frequency, and image rejection can be implemented. It is the dominant receiver architecture for radio, TV, and wireless communications due to its flexibility and performance.

Bit Error Rate (BER) is the ratio of erroneously received bits to total transmitted bits, indicating digital communication quality. BER = Number of bit errors / Total bits transmitted. Typical acceptable BER: 10^-6 for voice, 10^-9 for data. BER depends on SNR, modulation scheme, and channel conditions. Measurement involves transmitting known patterns (pseudo-random sequences) and comparing received data. BER curves (BER vs Eb/No) characterize system performance and are used for link budget calculations.

PCM is a method for digitizing analog signals through three steps: Sampling (measuring signal at regular intervals at Nyquist rate or higher), Quantization (mapping samples to discrete levels, introducing quantization noise), and Encoding (converting quantized values to binary). Standard telephone PCM uses 8kHz sampling and 8-bit quantization for 64 kbps. PCM enables digital transmission advantages: noise immunity, regeneration, easy multiplexing, and digital processing. Companding (A-law, mu-law) improves dynamic range for voice signals.

Optical fiber advantages include: Enormous bandwidth (terabits per second capacity), Low attenuation (0.2 dB/km for single-mode fiber, enabling long distances without repeaters), Immunity to electromagnetic interference (no crosstalk, secure from eavesdropping), Small size and light weight, Low cost (silica is abundant), and No ground loops or electrical hazards. Disadvantages include installation complexity, need for precise connectors, and difficulty with tight bends. Fiber dominates long-haul and backbone networks.

Noise types include: Thermal noise (Johnson noise, from random electron motion in conductors, white spectrum), Shot noise (from discrete charge carriers in semiconductors), Flicker noise (1/f noise, dominant at low frequencies), Intermodulation noise (from nonlinear mixing of signals), Crosstalk (interference from adjacent channels), and Atmospheric/cosmic noise (external interference). Thermal noise power = kTB (k=Boltzmann constant, T=temperature, B=bandwidth). System noise figure quantifies degradation from ideal performance.

Parity check adds one bit to data to make the total number of 1s either even (even parity) or odd (odd parity). The receiver checks if received parity matches expected; mismatch indicates error. Simple parity detects single-bit errors but cannot detect even numbers of errors or locate error position. Two-dimensional parity (adding row and column parities) can detect and correct single-bit errors. While simple, parity is used in memory systems (ECC DRAM uses more sophisticated codes) and basic error detection applications.

QPSK (Quadrature Phase Shift Keying) uses four phase states (typically 45, 135, 225, 315 degrees) to encode 2 bits per symbol. The constellation diagram shows these four points equally spaced on a circle. Advantages: doubled spectral efficiency over BPSK (2 bits/symbol), constant envelope suitable for nonlinear amplifiers. Implementation uses I and Q channels modulated by separate bit streams. QPSK has 3dB SNR penalty compared to BPSK for same BER. Variants include offset-QPSK and pi/4-QPSK for reduced envelope variations.

Shannon's theorem states maximum data rate C = B*log2(1 + SNR) bits/second, where B is bandwidth in Hz and SNR is signal-to-noise ratio. Key implications: There is a fundamental limit to error-free transmission rate, increasing bandwidth or SNR increases capacity, capacity approaches limit as SNR increases (diminishing returns), and practical systems operate below this limit. The theorem guides system design trade-offs between bandwidth, power, and complexity. Modern systems like LDPC and turbo codes approach Shannon limit within 1dB.

AM modulation index m = Em/Ec (message amplitude / carrier amplitude), also expressed as percentage. For m < 1: under-modulation, envelope follows message without distortion. For m = 1: 100% modulation, maximum efficiency without distortion. For m > 1: over-modulation causes envelope clipping and harmonic distortion. Power efficiency = m^2/(2+m^2), reaching 33% at m=1 for single-tone modulation. Sideband power is Psb = (m^2/4)*Pc. Proper modulation index balances power efficiency with distortion avoidance.

Frequency deviation (delta_f) is the maximum shift of carrier frequency from its center, determined by message amplitude. Modulation index beta = delta_f / fm (deviation / message frequency). FM bandwidth approximated by Carson's rule: BW = 2*(delta_f + fm) = 2*fm*(beta + 1). Narrowband FM (beta << 1) has BW similar to AM; wideband FM (beta > 1) has significantly larger bandwidth but better noise performance. Commercial FM broadcasting uses 75kHz deviation with 200kHz channel spacing.

QAM (Quadrature Amplitude Modulation) combines amplitude and phase modulation to encode multiple bits per symbol. 16-QAM uses 16 constellation points encoding 4 bits/symbol; 64-QAM uses 64 points for 6 bits/symbol; 256-QAM achieves 8 bits/symbol. Higher-order QAM increases spectral efficiency (bits/Hz) but requires higher SNR for same BER as constellation points become closer. Implementation requires precise amplitude control and linear amplifiers. QAM is used in cable modems, Wi-Fi, and LTE where good SNR is available.

Link budget accounts for all gains and losses in a communication link to ensure adequate received signal. Calculation: Pr = Pt + Gt - Lp + Gr - Lother, where Pt is transmit power, Gt/Gr are antenna gains, Lp is path loss (free space: 20*log10(4*pi*d*f/c)), and Lother includes cable losses, atmospheric effects, and margins. Link margin = Pr - receiver sensitivity. Design must ensure positive margin under worst-case conditions. Link budgets are essential for system design, determining required power, antenna sizes, and maximum range.

Convolutional codes generate output bits from current and previous input bits using shift registers and XOR operations. Described by (n,k,K) where n=output bits, k=input bits, K=constraint length. Code rate R=k/n. Unlike block codes, they operate on continuous streams. Decoding uses Viterbi algorithm (maximum likelihood, finds path through trellis with minimum distance to received sequence). Error correction capability depends on free distance. Used in deep-space communication, GSM, and 802.11a/g. Combined with interleaving for burst error correction.

Fading is signal strength variation due to multipath propagation. Large-scale fading: Path loss (distance-dependent attenuation) and Shadowing (slow fading from obstacles, log-normal distribution). Small-scale fading: Multipath fading (fast amplitude variations from constructive/destructive interference). Multipath is characterized as: Flat fading (bandwidth < coherence bandwidth) or Frequency-selective fading, Fast fading (symbol duration > coherence time) or Slow fading. Rayleigh fading models no line-of-sight; Rician fading includes dominant path. Mitigation techniques include diversity, equalization, and OFDM.

Companding (compressing + expanding) applies nonlinear quantization in PCM to improve dynamic range for signals with large amplitude variations. At transmitter, compression reduces amplitude range (small signals amplified more than large ones). At receiver, expansion reverses the process. A-law (Europe) and mu-law (North America, Japan) are standard companding laws. Benefits: Uniform quantization noise relative to signal level, improved SNR for quiet signals, reduced number of bits needed. 8-bit companded PCM achieves quality similar to 12-14 bit linear PCM.

In superheterodyne receiver with IF frequency fIF, when receiving signal at fRF with LO at fLO = fRF - fIF, an undesired image frequency fimage = fLO + fIF = fRF + 2*fIF also mixes to IF. Image rejection methods: Front-end RF filter (tunable bandpass, difficult at low IF), Image-reject mixer (Hartley or Weaver architecture using phase relationships), and High IF selection (places image further from desired signal, easier to filter but IF processing harder). Image rejection ratio is critical specification for receivers.

OFDM (Orthogonal Frequency Division Multiplexing) divides wideband channel into many narrowband subcarriers, each carrying low-rate data. Subcarriers are mathematically orthogonal, allowing overlap without interference, maximizing spectral efficiency. Implementation uses FFT/IFFT. Advantages: Robust against frequency-selective fading (each subcarrier sees flat fading), simple equalization (one-tap per subcarrier), efficient spectrum use, and easy implementation of MIMO. Cyclic prefix handles multipath. Used in Wi-Fi, LTE, DVB-T, and 5G NR.

AGC automatically adjusts receiver gain to maintain constant output level despite varying input signal strength. Components include: signal level detector, reference comparator, and variable gain amplifier. AGC is important because: received signal can vary over 100dB range (near/far problem), prevents overload distortion with strong signals, maintains optimal SNR with weak signals, and keeps subsequent stages operating in linear range. AGC loop bandwidth trades off response speed versus noise immunity. Fast AGC handles rapid fading; slow AGC for stable channels.

Interleaving rearranges data before transmission so that consecutive bits become separated. If burst error corrupts consecutive transmitted bits, de-interleaving at receiver spreads these errors among different code words. This converts burst errors into distributed random errors that error-correcting codes can handle. Types: Block interleaving (write row-wise, read column-wise) and Convolutional interleaving (variable delays). Interleaving depth depends on expected burst length. Trade-offs include latency (deeper interleaving adds delay) and memory requirements. Used in wireless, storage, and CD/DVD systems.

Spread spectrum spreads signal over bandwidth much larger than minimum required. DSSS (Direct Sequence Spread Spectrum): multiplies data by high-rate pseudo-random code, spreading bandwidth. FHSS (Frequency Hopping Spread Spectrum): rapidly switches carrier among frequencies in pseudo-random sequence. Benefits: Resistance to narrowband interference, LPI/LPD (low probability of intercept/detection), multiple access (CDMA), and multipath resistance. Applications: GPS (DSSS), Bluetooth (FHSS), military communications, and CDMA cellular. Processing gain = BW_spread / BW_signal improves SNR.

Dispersion causes pulse broadening, limiting fiber bandwidth. Types: Modal dispersion (multimode fiber only, different modes travel different paths), Chromatic dispersion (wavelength-dependent velocity, both material and waveguide components), and Polarization Mode Dispersion (PMD, random birefringence in fiber). Mitigation: Single-mode fiber eliminates modal dispersion, Dispersion-shifted fiber or compensation modules address chromatic dispersion, and PMD compensation for very high rates. Dispersion limits bandwidth-distance product and affects system reach at high data rates.

SSB modulation transmits only one sideband of AM signal (upper or lower), removing carrier and opposite sideband. Generation methods: Filter method (sharp sideband filter), Phase shift method (Weaver or Hartley), or Adaptive filters. Advantages: Bandwidth efficiency (half of conventional AM), Power efficiency (no carrier power wasted), Reduced fading (no carrier nulling). Disadvantages: Requires coherent demodulation (stable reference), more complex implementation. Used in amateur radio, HF communication, and professional radio where bandwidth and power are limited.

A matched filter maximizes output SNR at sampling instant for a known signal in additive white Gaussian noise. Its impulse response is time-reversed and delayed version of expected signal. For rectangular pulse s(t), matched filter is integrator sampled at pulse end. Mathematically, it correlates received signal with expected waveform. The matched filter is optimal because it concentrates signal energy at sampling instant while spreading noise. Used in radar, digital communications, and pattern detection. Implementation: FIR filter matched to pulse shape.

ARQ (Automatic Repeat reQuest) protocols use acknowledgments and retransmissions for reliability. Stop-and-Wait: Sender waits for ACK before sending next frame; simple but inefficient. Go-Back-N: Sender can have multiple outstanding frames, retransmits from error point; better efficiency. Selective Repeat: Only retransmits errored frames; most efficient but complex receiver buffering. ARQ combined with FEC creates Hybrid ARQ (HARQ): Type I discards errored packets, Type II/III combine soft information from retransmissions. Trade-offs: throughput, latency, complexity, and buffer requirements.

Direct conversion receiver directly converts RF to baseband without intermediate frequency, mixing with LO at carrier frequency. I/Q demodulation produces in-phase and quadrature baseband signals. Advantages: No image frequency problem, no IF filters, lower component count, easier integration. Challenges: DC offset (LO-to-RF leakage), I/Q imbalance, 1/f noise at baseband, LO pulling, and second-order distortion. Widely used in modern wireless (Wi-Fi, cellular) due to integration advantages. Requires calibration techniques for DC offset and I/Q balance.

CRC treats data as polynomial and divides by generator polynomial, appending remainder as check bits. At receiver, division of received data (including CRC) by same generator should yield zero remainder if no errors. CRC detects all single-bit errors, all burst errors shorter than CRC length, most longer bursts, and all odd-number bit errors (with factor x+1 in generator). Common CRCs: CRC-16, CRC-32 (Ethernet, ZIP). Implementation uses shift registers with XOR feedback matching generator polynomial. CRC is not suitable for correction, only detection.

MIMO (Multiple-Input Multiple-Output) uses multiple antennas at transmitter and receiver. Channel capacity scales with min(Nt, Nr) in rich scattering: C = log2(det(I + (SNR/Nt)*H*H^H)) where H is channel matrix. Spatial multiplexing transmits independent data streams on different spatial modes (eigenmodes of channel), achieving multiplexing gain. Requires channel knowledge for precoding (SVD decomposition) or spatial separation (V-BLAST detection). Trade-off between multiplexing gain (rate increase) and diversity gain (reliability). Massive MIMO scales to hundreds of antennas for 5G capacity improvement.

Turbo codes use parallel concatenation of two recursive systematic convolutional (RSC) encoders separated by an interleaver. Key innovations: Interleaver decorrelates inputs to second encoder, enabling iterative decoding; Soft-output decoders (BCJR algorithm) exchange extrinsic information between component decoders; Iterations refine reliability estimates, approaching MAP detection. Design considerations: Interleaver design (random, S-random), puncturing for rate adjustment, and convergence analysis (EXIT charts). Turbo codes come within 0.5dB of Shannon limit at moderate block lengths. Used in 3G, 4G, and deep-space communication.

OFDM's high Peak-to-Average Power Ratio (PAPR) occurs when subcarrier phases align, creating large peaks. High PAPR requires linear amplifiers with large back-off, reducing power efficiency. Reduction techniques: Clipping (simple but causes distortion and spectral regrowth), Coding (select codewords with low PAPR, reduces rate), Selected Mapping (SLM, transmit sequence with lowest PAPR from multiple phase rotations), Partial Transmit Sequences (PTS, optimize phase of subcarrier groups), Tone reservation (reserve subcarriers for peak reduction), and DFT-spreading (SC-FDMA in LTE uplink). Trade-offs involve complexity, rate loss, and distortion.

Coherent optical communication mixes received signal with local oscillator laser before detection, preserving phase and enabling complex modulation formats. Advantages: Improved receiver sensitivity (approaching quantum limit), Access to full optical field (amplitude and phase), enabling PM-QPSK, 16-QAM etc., Linear channel response enables digital signal processing for impairment compensation, and Flexible wavelength selection without optical filters. DSP algorithms compensate chromatic dispersion, PMD, and laser phase noise digitally. Enables 100G/400G per wavelength transmission over long distances. Challenges include laser linewidth requirements and DSP complexity.

LDPC (Low-Density Parity-Check) codes are linear block codes defined by sparse parity-check matrix H. Design: Regular codes have fixed column/row weights; irregular codes optimize degree distribution for better threshold (using density evolution or EXIT chart analysis). Matrix construction: Random (capacity-approaching but high floor), structured (QC-LDPC for efficient implementation). Decoding: Belief propagation (sum-product algorithm) iteratively passes messages between variable and check nodes. Implementation uses min-sum approximation for complexity reduction. LDPC achieves within 0.1dB of Shannon limit and is used in DVB-S2, Wi-Fi 6, and 5G NR.

Equalizers compensate inter-symbol interference (ISI) from multipath. Types: Linear equalizers (ZF forces zero ISI but enhances noise, MMSE balances ISI and noise), Decision feedback (DFE uses decisions to cancel post-cursor ISI, sensitive to error propagation), and Maximum likelihood sequence estimation (MLSE via Viterbi, optimal but exponential complexity with channel length). Adaptive algorithms: LMS (simple, slow convergence), RLS (fast, higher complexity). OFDM uses frequency-domain equalization (one-tap per subcarrier). Design considerations: Training sequence for channel estimation, tracking speed, complexity, and interaction with FEC.

Carrier synchronization recovers carrier frequency and phase for coherent demodulation. Methods: Squaring loop (for BPSK, squares signal to remove modulation, PLL locks to 2x carrier), Costas loop (parallel I/Q processing, generates phase error independent of data), Decision-directed (uses symbol decisions to generate error signal, works after timing recovery). For suppressed carrier (PSK), these must be phase-ambiguity aware. Frequency acquisition uses frequency detector (at large errors) then phase detector. Loop bandwidth trades acquisition speed vs phase noise. Digital implementations use NCO and loop filters with configurable bandwidth.

Diversity combines multiple independently fading signal copies to improve reliability. Techniques: Selection combining (selects branch with highest SNR, simplest, diversity gain d=N), Equal gain combining (coherently sums equal-weighted branches, 1dB less than MRC), Maximal ratio combining (MRC, weights proportional to SNR, optimal, maximizes output SNR). For Rayleigh fading, MRC with N branches achieves BER improvement of SNR^(-N). Implementation: Requires channel estimation for MRC weights, selection needs only branch SNR monitoring. Space, time, frequency, and polarization domains provide diversity. Coding also provides diversity through redundancy across fades.

Timing synchronization determines optimal sampling instant. Methods: Early-late gate (compares samples early/late relative to decision point, simple), Gardner algorithm (timing error from difference of samples around zero crossing, self-timing), Mueller and Muller (uses decisions, works with bandwidth-efficient pulses). For OFDM: Cyclic prefix correlation for coarse sync, pilot-based fine timing. Implementation: Timing error detector drives loop filter controlling interpolator or ADC clock. Interpolation architectures: Polyphase (explicit filters), Farrow (polynomial), and NCO-controlled. Timing jitter directly impacts BER through ISI.

Network information theory extends Shannon's work to multi-user scenarios. Key results: Multiple access channel (MAC) capacity region defines achievable rate pairs for multiple transmitters to one receiver, Broadcast channel capacity for one transmitter to multiple receivers, Relay channel strategies (decode-forward, compress-forward, amplify-forward), Interference channel (capacity unknown in general, strategies include treating interference as noise, decoding interference, Han-Kobayashi). Practical implications: NOMA (non-orthogonal multiple access), cooperative communications, and network coding. Capacity achieving often requires sophisticated coding across users.

SDR implements radio functionality in software for flexibility. Architecture considerations: ADC/DAC requirements (sampling rate, resolution, spurious-free dynamic range), RF front-end (wideband or channelized, noise figure, linearity), Processing platform (FPGA for real-time PHY, GPP for MAC and higher layers, or hybrid), Memory bandwidth and latency, and Synchronization between processing stages. Key challenges: Processing throughput for high bandwidth signals, latency for real-time applications, power consumption for mobile, and achieving performance comparable to dedicated hardware. SDR enables cognitive radio, multi-standard operation, and rapid prototyping.

WDM (Wavelength Division Multiplexing) transmits multiple wavelengths on single fiber. Design considerations: Channel spacing (100GHz, 50GHz, or flex-grid DWDM), wavelength plan and guard bands, amplifier design (EDFA gain flatness across band, noise figure), fiber nonlinearity management (self-phase modulation, cross-phase modulation, four-wave mixing), dispersion map design, OSNR requirements per channel, and amplifier spacing. Advanced systems: Coherent detection enables higher spectral efficiency, flex-grid allows varied channel widths, and Raman amplification extends reach. Capacity exceeds 10 Tbps per fiber in modern systems.

Polar codes provably achieve channel capacity through channel polarization: recursively combining channels creates extremes of reliable and unreliable bit channels. Information bits use reliable channels; frozen bits (known values) use unreliable ones. Encoding is simple (multiplication by generator matrix, O(N log N)). Decoding: Successive cancellation (SC, O(N log N)), SC-List (near-ML performance), or belief propagation. 5G selection rationale: excellent performance at short block lengths (control channels), low complexity, proven capacity-achieving, and efficient encoding. Polar codes are first provably capacity-achieving codes with practical complexity.

Cognitive radio dynamically accesses unused spectrum (white spaces) without interfering with primary users. Spectrum sensing methods: Energy detection (compares received power to threshold, simple but susceptible to noise uncertainty), Feature detection (exploits signal characteristics like cyclostationarity, pilots), Matched filter (optimal for known signals), Cooperative sensing (multiple CRs share decisions for improved reliability against shadowing). Challenges: Detecting weak primary signals, hidden node problem, sensing-throughput trade-off, and fast detection under mobility. Database approaches complement sensing by querying spectrum availability from regulatory database.

Full-duplex transmits and receives simultaneously on same frequency, potentially doubling spectral efficiency. Main challenge: Self-interference (SI) from transmitter to receiver can be 100+ dB stronger than desired signal. Cancellation stages: Antenna isolation (separation, directional antennas, circulators, 15-40dB), RF cancellation (inject inverted TX signal, 20-40dB), and Digital cancellation (estimate and subtract residual SI, 30-50dB). Challenges: Hardware nonlinearities and noise limit achievable cancellation, and receiver chain must handle large SI without saturation. Applications: Relaying, wireless backhaul, and collision detection. Practical implementations achieve total 110dB+ cancellation.