Finite Fields and Confidentiality · Cryptography and Network Security

Traffic Confidentiality

Watch lectures

Traffic Confidentiality

1. Beyond Content: Hiding Who Talks and How Much

Encryption hides WHAT is said; traffic confidentiality hides communication patterns — source/destination identities, frequency, length, timing — which alone reveal plenty (embassy–capital spike before an announcement; fixed-size hourly bursts = telemetry). Stallings's measures escalate: (a) link encryption (hides headers per hop), (b) padding, (c) dummy traffic, (d) full covertness via mixes/onion routing.

Attacker sees: (src,dst,ti,leni)⇒infer relationship + activity even with C=EK(P) unbroken\text{Attacker sees: } (src, dst, t_i, len_i) \Rightarrow \text{infer relationship + activity even with } C = E_K(P) \text{ unbroken}

2. Techniques (Stallings List + Modern Forms)

Technique How What it hides Cost
Link encryption Encrypt whole frame per hop Addresses + payload on wire Per-hop decrypt (trust nodes)
Padding / constant rate Pad to fixed size / fixed schedule Length + timing signatures Bandwidth waste
Dummy / cover traffic Random-filler packets when idle Existence of real bursts High overhead (military-grade)
Mixes (Chaum) Batch + reorder + re-encrypt Input↔output linkage Latency (remailers, voting)
Onion routing (Tor) Layered encrypt, 3-hop circuit Source from destination and vice versa Latency, exit-node trust
ECH / encrypted SNI Encrypt TLS handshake server name Destination hostname from snoopers Needs DNS-over-HTTPS + server support

Worked padding math: messages len∈{100,500,900}len \in \{100, 500, 900\} → pad all to 1024: attacker learns only count, not which; overhead =(1024−lenˉ)×N= (1024 - \bar{len}) \times N. Constant-rate 1 packet/s with dummies when idle: timing signal fully flat; cost = peak-rate bandwidth always.

3. What TLS/IPsec Leak (Exam Favourites)

  • TLS ≤1.2: SNI hostname, cert, packet sizes/timing visible — website fingerprinting identifies pages from size sequences (~90% accuracy in lab).
  • IPsec tunnel: inner addresses hidden, but outer endpoints + counts/sizes visible — flow correlation possible.
  • Metadata programs (and BEU case studies): "we only collected metadata" still maps social graphs, sleep patterns, business deals — hence ECH, QUIC padding, Tor.

4. Limits and Design Rule

Perfect traffic confidentiality = constant-rate cover traffic on every link — prohibitively expensive; real systems buy levels: ECH + padding (cheap, HTTPS), VPN/Tor (moderate, metadata-hiding), constant-rate + mixes (expensive, high-threat). Rule: classify what patterns hurt (identity? timing? volume?), then pay only for those — padding sizes is cheap, faking timing is dear.

Exam Points

  • Content-vs-pattern distinction + (src,dst,t,len)(src, dst, t, len) tuple — 2-mark opener.
  • Six-technique table with hides/cost columns — 5-mark core.
  • Padding math ({100,500,900}→1024\{100,500,900\} \to 1024) + constant-rate dummy logic.
  • TLS/IPsec leak list (SNI, sizes, outer endpoints) + "metadata maps graphs" closer.
Official AppAndroid App

PYQDeck for Android

Solve BEU previous-year exam papers anywhere, anytime. Fast search, offline paper viewing, and step-by-step solutions designed specifically for your phone.

← All Cryptography and Network Security modules

Traffic Confidentiality — Cryptography and Network Security — PYQDeck