Check: Ad-Hoc Readiness Scenarios and Failover

Knowledge check · self-test
Check: Ad-Hoc Readiness Scenarios and Failover
Disaster Multi-Hop Range Planning. Scenario: Earthquake destroys cellular infrastructure. 50 rescue workers need communication across 2km disaster zone. Each radio: 500m range, 1W transmit power, battery lasts 8 hours continuous transmission. Think about: Calculate minimum hops for end-to-end connectivity (2000m coverage). Why not use a single 2 km-range radio (16× transmit power in this free-space example for the same received-power target)? ITU-R P.525-5. How does multi-hop extend battery life in energy-constrained scenarios? Key Insight: Minimum hops = 2000m / 500m = 4 hops. Path: Source -> Node1(500m) -> Node2(1000m) -> Node3(1500m) -> Dest(2000m). In a free-space link model, received power falls with the square of distance; doubling the distance requires four times the transmit power to hold received power constant. ITU-R P.525-5 Under that free-space, equal-antenna link model, a single 2 km link needs 16× the transmit power of a 500 m link (1 W × 16 = 16 W in this example), compared with four 500 m hops transmitting at 1 W each. ITU-R P.525-5 Battery impact: 16W transmission = 30 minutes runtime vs 1W multi-hop = 8 hours. Multi-hop isn't just about range - it's about energy efficiency. Disaster response networks prioritize battery life over latency. Social Routing Efficiency. Scenario: University campus deploys delay-tolerant messaging app for 5,000 students. Epidemic routing floods 50+ message copies per message. SocialCast analyzes: contact frequency (Alice-Bob meet daily), recency (last contact 2 hours ago), location similarity (both frequent library). Think about: Why does epidemic routing work but waste bandwidth? In this illustrative campus scenario, how would SocialCast reduce copies from 50 to 5–10 while targeting 90% delivery? What happens when social patterns are unpredictable (random mobility)? Key Insight: In this campus example, epidemic routing floods copies broadly, but it does not guarantee delivery; compare the assumed copy counts before claiming bandwidth savings. RFC 4838 SocialCast = forward only if carrier's social metrics > current holder's. Example: Message for "Bob". Current holder "Charlie" (meets Bob monthly). Encounters "Alice" (meets Bob daily) -> transfer to Alice (higher contact frequency). Encounters "Dave" (never meets Bob) -> don't transfer. Result: 5-10 targeted copies follow social graph toward Bob. In this illustrative campus scenario, assume 90% delivery and 5–10 targeted copies instead of 50; measure transmitted bytes before claiming bandwidth savings. Perfect for human mobility (predictable social patterns). Fails for random mobility (no patterns to exploit) - use epidemic instead. Adaptive Route Caching. Scenario: Disaster response team uses DSR routing. Command post (stationary) communicates with mobile rescue teams (3-5 m/s). Device caches 5 routes. After 10 minutes, 40% of routes have stale/broken links due to mobility. Think about: Calculate optimal cache timeout balancing freshness vs discovery overhead. Why not always perform fresh route discovery (zero cache)? How does mobility speed affect ideal timeout? Key Insight: Static timeout fails: 10min timeout -> 40% stale routes -> packet failures. 30sec timeout -> excessive RREQ floods (high overhead). Adaptive solution: Monitor ROUTE_ERROR rate: High errors (>5/min) -> reduce timeout to 2-3min (high mobility). Low errors ( extend timeout to 10min (stable topology). Mobility-aware: Command post to command post: 10min cache (stationary). Rescue team to rescue team: 2min cache (mobile). Mixed: 5min cache. Why not zero cache? Discovery latency: 50-200 ms per RREQ/RREP cycle. Real-time voice needs <100 ms. Cached routes = instant forwarding. Result: Adaptive caching = DSR's low latency + mobility resilience. Route lifetime inversely proportional to node velocity. Energy-Aware Multi-Path Selection. Scenario: 20-node ad-hoc network has 3 paths from source to gateway: Path A (3 hops, nodes at 40% battery), Path B (4 hops, nodes at 80% battery), Path C (5 hops, nodes at 95% battery). Need to route 1000 packets while maximizing network lifetime. Think about: Why doesn't shortest-path routing (Path A) maximize network lifetime? Calculate battery depletion if all traffic uses Path A vs distributed across all paths. How does energy-aware routing trade latency for longevity? Key Insight: Shortest-path (Path A only): 1000 packets drain 40% -> 0% battery on 3 nodes -> network partition (other nodes isolated). Energy-aware distribution: 300 packets on A (40% → 28%), 400 packets on B (80% -> 70%), 300 packets on C (95% -> 88%). Minimum battery: 28% (vs 0% shortest-path). Spreading this scenario’s 1,000 packets across three paths leaves the weakest path at 28% battery rather than 0%; estimating lifetime also needs a time and failure model. Trade-offs: Under an equal per-hop-delay assumption, Path C has 67% higher path delay (5 hops versus 3) but uses nodes with abundant energy. Production strategy: Route delay-tolerant traffic on long/high-battery paths, route real-time traffic on short/low-latency paths. Prevents premature node deaths that partition network. Mobile Link Quality Prediction. Scenario: Ad-hoc network link between Node A and Node B: current RSSI -65 dBm, PDR 95%, nodes approaching at 2 m/s, will separate in 30 seconds. Link quality monitor must predict when link becomes unusable (RSSI < -85 dBm, PDR < 50%). Think about: Calculate when proactive rerouting should trigger (before link failure). Why is prediction better than reactive rerouting after failure? How does mobility speed affect prediction accuracy? Key Insight: Path loss increases with distance: RSSI drops ~3-4 dB per doubling distance. At 2 m/s separation velocity: Distance = 2 m/s x 30s = 60m additional separation. RSSI degrades: -65 dBm (now) -> -75 dBm (15s) -> -85 dBm (30s) -> link fails. Proactive rerouting: Trigger new route discovery at 20s (RSSI -80 dBm, PDR 70%) before complete failure. Gives 10s buffer for RREQ/RREP exchange. Reactive rerouting: Wait until link fails (PDR buffer loss, 200 ms discovery delay, user-visible disruption. Prediction accuracy: High mobility (5 m/s+) -> prediction horizon 10-15s (short). Low mobility (1 m/s) -> horizon 60s+ (plenty of time). Use velocity-aware prediction windows.
Disaster Multi-Hop Range Planning. Scenario: Earthquake destroys cellular infrastructure. 50 rescue workers need communication across 2km disaster zone. Each radio: 500m range, 1W transmit power, battery lasts 8 hours continuous transmission. Think about: Calculate minimum hops for end-to-end connectivity (2000m coverage). Why not use a single 2 km-range radio (16× transmit power in this free-space example for the same received-power target)? ITU-R P.525-5. How does multi-hop extend battery life in energy-constrained scenarios? Key Insight: Minimum hops = 2000m / 500m = 4 hops. Path: Source -> Node1(500m) -> Node2(1000m) -> Node3(1500m) -> Dest(2000m). In a free-space link model, received power falls with the square of distance; doubling the distance requires four times the transmit power to hold received power constant. ITU-R P.525-5 Under that free-space, equal-antenna link model, a single 2 km link needs 16× the transmit power of a 500 m link (1 W × 16 = 16 W in this example), compared with four 500 m hops transmitting at 1 W each. ITU-R P.525-5 Battery impact: 16W transmission = 30 minutes runtime vs 1W multi-hop = 8 hours. Multi-hop isn't just about range - it's about energy efficiency. Disaster response networks prioritize battery life over latency. Social Routing Efficiency. Scenario: University campus deploys delay-tolerant messaging app for 5,000 students. Epidemic routing floods 50+ message copies per message. SocialCast analyzes: contact frequency (Alice-Bob meet daily), recency (last contact 2 hours ago), location similarity (both frequent library). Think about: Why does epidemic routing work but waste bandwidth? In this illustrative campus scenario, how would SocialCast reduce copies from 50 to 5–10 while targeting 90% delivery? What happens when social patterns are unpredictable (random mobility)? Key Insight: In this campus example, epidemic routing floods copies broadly, but it does not guarantee delivery; compare the assumed copy counts before claiming bandwidth savings. RFC 4838 SocialCast = forward only if carrier's social metrics > current holder's. Example: Message for "Bob". Current holder "Charlie" (meets Bob monthly). Encounters "Alice" (meets Bob daily) -> transfer to Alice (higher contact frequency). Encounters "Dave" (never meets Bob) -> don't transfer. Result: 5-10 targeted copies follow social graph toward Bob. In this illustrative campus scenario, assume 90% delivery and 5–10 targeted copies instead of 50; measure transmitted bytes before claiming bandwidth savings. Perfect for human mobility (predictable social patterns). Fails for random mobility (no patterns to exploit) - use epidemic instead. Adaptive Route Caching. Scenario: Disaster response team uses DSR routing. Command post (stationary) communicates with mobile rescue teams (3-5 m/s). Device caches 5 routes. After 10 minutes, 40% of routes have stale/broken links due to mobility. Think about: Calculate optimal cache timeout balancing freshness vs discovery overhead. Why not always perform fresh route discovery (zero cache)? How does mobility speed affect ideal timeout? Key Insight: Static timeout fails: 10min timeout -> 40% stale routes -> packet failures. 30sec timeout -> excessive RREQ floods (high overhead). Adaptive solution: Monitor ROUTE_ERROR rate: High errors (>5/min) -> reduce timeout to 2-3min (high mobility). Low errors ( extend timeout to 10min (stable topology). Mobility-aware: Command post to command post: 10min cache (stationary). Rescue team to rescue team: 2min cache (mobile). Mixed: 5min cache. Why not zero cache? Discovery latency: 50-200 ms per RREQ/RREP cycle. Real-time voice needs <100 ms. Cached routes = instant forwarding. Result: Adaptive caching = DSR's low latency + mobility resilience. Route lifetime inversely proportional to node velocity. Energy-Aware Multi-Path Selection. Scenario: 20-node ad-hoc network has 3 paths from source to gateway: Path A (3 hops, nodes at 40% battery), Path B (4 hops, nodes at 80% battery), Path C (5 hops, nodes at 95% battery). Need to route 1000 packets while maximizing network lifetime. Think about: Why doesn't shortest-path routing (Path A) maximize network lifetime? Calculate battery depletion if all traffic uses Path A vs distributed across all paths. How does energy-aware routing trade latency for longevity? Key Insight: Shortest-path (Path A only): 1000 packets drain 40% -> 0% battery on 3 nodes -> network partition (other nodes isolated). Energy-aware distribution: 300 packets on A (40% → 28%), 400 packets on B (80% -> 70%), 300 packets on C (95% -> 88%). Minimum battery: 28% (vs 0% shortest-path). Spreading this scenario’s 1,000 packets across three paths leaves the weakest path at 28% battery rather than 0%; estimating lifetime also needs a time and failure model. Trade-offs: Under an equal per-hop-delay assumption, Path C has 67% higher path delay (5 hops versus 3) but uses nodes with abundant energy. Production strategy: Route delay-tolerant traffic on long/high-battery paths, route real-time traffic on short/low-latency paths. Prevents premature node deaths that partition network. Mobile Link Quality Prediction. Scenario: Ad-hoc network link between Node A and Node B: current RSSI -65 dBm, PDR 95%, nodes approaching at 2 m/s, will separate in 30 seconds. Link quality monitor must predict when link becomes unusable (RSSI < -85 dBm, PDR < 50%). Think about: Calculate when proactive rerouting should trigger (before link failure). Why is prediction better than reactive rerouting after failure? How does mobility speed affect prediction accuracy? Key Insight: Path loss increases with distance: RSSI drops ~3-4 dB per doubling distance. At 2 m/s separation velocity: Distance = 2 m/s x 30s = 60m additional separation. RSSI degrades: -65 dBm (now) -> -75 dBm (15s) -> -85 dBm (30s) -> link fails. Proactive rerouting: Trigger new route discovery at 20s (RSSI -80 dBm, PDR 70%) before complete failure. Gives 10s buffer for RREQ/RREP exchange. Reactive rerouting: Wait until link fails (PDR buffer loss, 200 ms discovery delay, user-visible disruption. Prediction accuracy: High mobility (5 m/s+) -> prediction horizon 10-15s (short). Low mobility (1 m/s) -> horizon 60s+ (plenty of time). Use velocity-aware prediction windows.
Back to Ad-Hoc Production: Readiness Checks: Transfer the Evidence Across Deployments