Mock Interview Mastery — 15 Questions, 30-40 Minutes

After your 15-20 minute presentation, the examiner opens the floor to 30-40 minutes of technical questioning. This is where 60% of your grade lives (Technical Depth 40% + Problem-Solving 20%).

The STAR-T Response Structure

For every interview answer, use this framework:

• S — Set up: Direct answer first (1-2 sentences). Don't build to the point — lead with it.
• T — Technical detail: Expand with specifics. Name the RFC, the class name, the parameter value.
• A — AETHERIX context: Tie it to YOUR project. "In AETHERIX, this is implemented as..."
• R — Rationale: Why you made this choice. "We chose X over Y because..."
• T — Trade-off: What's the limitation or alternative? "The downside is..."

Example — Q: "Why DTN over TCP?"

(S) TCP is unsuitable for space because it assumes low latency, continuous connectivity, and end-to-end acknowledgements. (T) At Mars's average RTT of 25 minutes, a TCP three-way handshake alone takes 37.5 minutes. (A) In AETHERIX, we use Bundle Protocol v7 (RFC 9171) with store-and-forward — each node holds bundles until the next contact window. (R) We chose this because contact windows are only 6-12 hours/day. (T) The trade-off is increased storage requirements at relay nodes, which we mitigate with data prioritization.

LO1: DTN Fundamentals

Q1: Why is TCP/IP unsuitable for interplanetary communication? (2 min) - Three TCP assumptions break: low latency, continuous connectivity, end-to-end ACKs - TCP handshake to Mars: 1.5 × 25 min RTT = 37.5 min just to connect - DTN solves with store-and-forward + custody transfer + convergence layers

Q2: Explain BPv7 bundle structure and priority classes. (2 min) - Primary block (CBOR): source/dest EID, timestamp, lifetime, hop count, flags - Payload block: application data - Extension blocks: security, metadata - 5 priority levels: P0 Emergency → P4 Bulk - Flags: CUSTODY_REQUESTED (0x08), IS_FRAGMENT (0x01)

Q3: How do convergence layers adapt to different link types? (2 min) - LTP (RFC 5326): deep space — red/green segments, checkpoint-based retransmission - TCPCL (RFC 7242): Earth segment — reliable, TCP-based - UDP-CL: inter-satellite laser links — high speed, loss modeling - Bundle layer sits ABOVE transport — swap convergence layers per hop

LO2: RL-Based Routing

Q4: Explain your RL agent's state, action, and reward design. (2.5 min) - State: 8 variables (link quality, buffer occupancy, bundle priority, time-to-deadline, neighbor count, contact schedule, energy level, congestion) - Actions: forward, store, drop, split - Reward: R = α·delivery(1.0) − β·delay − γ·hops − δ·drops(10.0) − ε·energy - Epsilon-greedy, decay 0.995, CGR fallback at confidence < 0.3

Q5: How does multi-agent federated learning work? (2 min) - Each node runs its own Q-learning agent - Periodically aggregate Q-tables (weighted average) - Preserves privacy — no raw data shared, only learned policy parameters - Faster convergence than single-agent on large networks

LO3: Quantum Communication

Q6: Walk through BB84. (2.5 min) - 8 steps: random bits → random bases → send qubits → Bob measures → compare bases → sift → QBER → security decision - QBER < 11% (Shor-Preskill) → secure - Eavesdropper introduces ~25% QBER → detected

Q7: Why both QKD AND post-quantum crypto? (2 min) - Defense-in-depth - QKD: information-theoretic security, but needs hardware, low key rate (1-10 bps Mars) - ML-KEM (FIPS 203): fallback key exchange over any channel - ML-DSA (FIPS 204): authentication — QKD can't authenticate - 3,309-byte signatures: 0.3% overhead on 1MB bundle, 33× on 100-byte command

LO4: Space Infrastructure

Q8: Explain your 5-tier topology. (2.5 min) - Tier 1: Earth Ground (DSN: Goldstone, Madrid, Canberra — 120° apart) - Tier 2: Earth Orbital (3 GEO + 48 LEO laser constellation, Walker Delta) - Tier 3: Deep Space (ES-L4, ES-L5 Lagrange relays + quantum repeaters) - Tier 4: Mars Orbital (areostationary at 17,032 km) - Tier 5: Mars Surface (167 nodes: bases, rovers, drones, sensors) - Total: 241 nodes

Q9: Derive the link budget for Earth-Mars optical. (3 min) - FSPL = 20·log₁₀(4πd/λ) - At 225M km, 1550nm: FSPL ≈ 365 dB - TX: 5W, 22cm aperture → EIRP calculation - RX: 1m telescope → received power ≈ -234 dBm - Result: 10-200 Mbps depending on distance

Q10: How do Lagrange relays maintain connectivity during conjunction? (2 min) - Solar conjunction: Sun blocks direct Earth-Mars line-of-sight for ~2 weeks - ES-L4 and ES-L5 are 60° ahead/behind Earth — they can "see around" the Sun - RF fallback via Lagrange relays: 50-70% conjunction availability - Optical blackout during peak conjunction — Ka-band RF maintains minimum data flow

LO5: Radiation & Data Priority

Q11: What radiation effects threaten spacecraft electronics? (2 min) - SEU: Single Event Upset (bit flip) - SEL: Single Event Latch-up (destructive current) - MBU: Multiple Bit Upset - TID: Total Ionizing Dose (cumulative degradation) - Mars: no global magnetic field → higher radiation than Earth orbit

Q12: How does your 4-tier priority system work? (2 min) - P0 Emergency: anomaly alerts, < 1 min - P1 Mission-critical: science events, < 30 min - P2 High-priority: daily science, < 24 hr - P3 Standard: housekeeping, < 7 days - P4 Bulk: archives, < 30 days - Emergency preemption: P0 jumps queue regardless of buffer state - CCSDS 121/122 compression reduces payload size before queuing

LO6: Standards & Design Rationale

Q13: Which standards does AETHERIX comply with? (2 min) - RFCs: 9171 (BPv7), 5326 (LTP), 7242 (TCPCL), 4838 (DTN Architecture) - CCSDS: 734.2-B-1 (DTN), RFC 9172 (BPSec security), 141.0-B-1 (Optical), 401.0-B-30 (RF), 121/122 (Compression) - NIST: FIPS 203 (ML-KEM), 204 (ML-DSA), 205 (SLH-DSA)

Q14: What was your most challenging design decision? (2 min) - RL routing vs pure CGR - Decision: hybrid approach — RL primary, CGR fallback at confidence < 0.3 - Rationale: RL discovers novel routes CGR can't, but CGR provides guaranteed baseline - Trade-off: RL requires training time and doesn't guarantee optimality

Q15: If you could add one feature, what would it be? (2 min) - Software-defined radio (SDR) for dynamic spectrum allocation - Or: machine learning for solar flare prediction to pre-emptively route around conjunctions - Or: blockchain for tamper-evident command logging - Pick ONE and defend it. Don't list three half-baked ideas.

1. Admit Uncertainty Gracefully

If you don't know something:

"That's a great question. I focused primarily on [X] in this project. For [Y], I'd need to research further, but my initial thinking would be [reasonable speculation]."

Never: "I don't know." (Dead end) Never: Make up an answer. (They'll catch it) Always: Pivot to what you DO know.

2. The "Yes, and..." Technique

When the examiner adds complexity:

Examiner: "What if two Lagrange relays fail simultaneously?" You: "That's an important edge case. With both ES-L4 and ES-L5 down, the primary conjunction path is severed. The system would fall back to..." [describe Mars-relay → direct-to-DSN when line-of-sight permits, accepting reduced availability during conjunction]

Acknowledge the scenario is valid, then reason through it.

3. Quantify Everything

Vague answers lose points. Specific numbers win: - ❌ "High data rate" → ✅ "100 Mbps at closest approach" - ❌ "Long delay" → ✅ "3-22 minutes one-way" - ❌ "Many nodes" → ✅ "241 nodes across 5 tiers" - ❌ "Good security" → ✅ "QBER < 11%, Shor-Preskill threshold"

4. Don't Over-Answer

2 minutes per question. If you finish in 90 seconds and covered the key points, STOP. Don't pad. Examiners appreciate conciseness — it shows mastery.

5. The Follow-Up Trap

Examiners often probe deeper after a good answer. This is a GOOD sign — they're engaged. But be careful: - If they push into areas beyond your project scope, say so: "That's beyond what AETHERIX simulates, but the theoretical framework would be..." - If they push on a weakness, acknowledge it honestly: "You're right, that's a limitation. The mitigation I implemented was..."

• Mock Interview Script
• Question Bank — 106 practice questions
• Topic Summaries — 11 topic deep-dives
• Cheat Sheets — formulas + constants

The final lesson. Everything comes together. Rest, preparation, mindset, and the ultimate cheat sheet.