|
| 1 | +# Computational, Authenticated Key Distribution |
| 2 | + |
| 3 | +Encryption, the old-fashioned way |
| 4 | + |
| 5 | +## Value statement |
| 6 | + |
| 7 | +- Computational cryptography secures communication using modest computational resources |
| 8 | +- It is deployed on a vast share of the worlds computational resources |
| 9 | +- From a practical standpoint, computational cryptography is superior to QKD in all use cases without exception |
| 10 | +- There are some features, computational cryptography can only provide by incorporating QKD |
| 11 | + |
| 12 | +Handshake graphic |
| 13 | + |
| 14 | +- Introduce the methods used in cryptography |
| 15 | +- Identify the value statement QKD can provide given the above constraints |
| 16 | +- Contrast QKD and computational cryptography |
| 17 | + |
| 18 | +## About me |
| 19 | + |
| 20 | +My picture |
| 21 | + |
| 22 | +- Karolin Varner |
| 23 | +- Spent a decade as a Software Engineer in industry working for Startups and Corporate Behemoths |
| 24 | +- Started a cryptography research project in 2021 |
| 25 | +- Researcher Engineer at Max Planck Institute for Security and Privacy |
| 26 | +- Research interests: Key exchange protocols, computer-aided proofs of security, decryption despite error |
| 27 | +- Main founder of Rosenpass e.V.; founded to support interdisciplinary work |
| 28 | + |
| 29 | +## Rosenpass |
| 30 | + |
| 31 | +… Rosenpass slide |
| 32 | ++ Interdisciplinary |
| 33 | + |
| 34 | +## Cryptography Science Workflow |
| 35 | + |
| 36 | +:::info |
| 37 | + |
| 38 | +@mullana |
| 39 | + |
| 40 | +This might be to complex as it is here in writing. |
| 41 | +I can create a template graphic for you. |
| 42 | + |
| 43 | + |
| 44 | +Swirls |
| 45 | + |
| 46 | +- Identify goals colloquially "Secure communication" |
| 47 | +- Operationalize goals "Authenticity, Secrecy" |
| 48 | +(Philosophical) |
| 49 | + |
| 50 | +- Formalize security notions "IND-CCA2, forward secrecy" |
| 51 | +- Compare contrast and compose "eCK-PFS-PSK" |
| 52 | +(Mathematical) |
| 53 | + |
| 54 | +- Develop mathematical construction (e.g. Noise-IK) |
| 55 | +- Create security proof |
| 56 | +(Mathematical) |
| 57 | + |
| 58 | +- Implement concrete protocol (e.g. WireGuard) |
| 59 | +- Optimize concrete protocol |
| 60 | +- Find & Attack implementation flaws |
| 61 | +(Computer science) |
| 62 | + |
| 63 | +- Analyze practical security concerns (e.g. side-channel attacks) |
| 64 | +- Analyze usability of implementation |
| 65 | +(Computer science, Electrical Engineering, Physics, etc.) |
| 66 | + |
| 67 | +- Disseminate |
| 68 | +- Deploy |
| 69 | +- Analyze practical security aspects |
| 70 | +- Improve usability |
| 71 | +(Operations, Psychology, Social sciences, Human interface design) |
| 72 | + |
| 73 | +- Analyze social impact |
| 74 | +(Social sciences) |
| 75 | + |
| 76 | +Lots of circles in one bigger circle spiraling out. |
| 77 | +Each subcirle has an arrow going out "novel attacks" |
| 78 | + |
| 79 | +::: |
| 80 | + |
| 81 | +## Math proofs are fundamental to cryptography |
| 82 | + |
| 83 | +Reduction to mathematical problem: |
| 84 | + |
| 85 | +1. Assume attacker (stateful function) against cryptographic system |
| 86 | +2. Construct algorithm that solves mathematical problem using assumed attacker |
| 87 | + |
| 88 | +Information theoretic. |
| 89 | + |
| 90 | +1. Formulate cryptosystems as a function where |
| 91 | + * K – key material – represents any secret information held by all parties |
| 92 | + * D – protected information – represents any information to be protected |
| 93 | + * C – leaked information – represents any information known to the attacker after execution of the cryptosystem |
| 94 | + |
| 95 | + $F : K \times D \to C$ |
| 96 | + |
| 97 | +2. Show that every every possible value of the leaked information, every possible protected information is equally likely |
| 98 | + |
| 99 | + $\forall c : C, d_1 : D, d_2 : D; |{ k \in K | F(k, d_1) = c }| = |{ k \in K | F(k, d_1) = c }|$ |
| 100 | + |
| 101 | +Functional correctness of implementations: |
| 102 | + |
| 103 | +1. Using formal methods from computer science to show that a cryptographic implementation is equivalent to its specification |
| 104 | + |
| 105 | +Implementation security: |
| 106 | + |
| 107 | +1. Using mechanized verification to show that an implementation fulfills security properties such as |
| 108 | + - Timing side-channel resistance (certain assembly operations are forbidden) |
| 109 | + - Memory-safety (utilize advanced programming languages such as Rust to avoid buffer overflow and other memory safety errors) |
| 110 | + |
| 111 | +Efficiency of implementation: |
| 112 | + |
| 113 | +2. Using complexity analysis to show that an implementation's resource usage is efficient-enough |
| 114 | + |
| 115 | +## Practical Security essential in cryptography |
| 116 | + |
| 117 | +- Protection against timing side-channels |
| 118 | +- Power side-channels |
| 119 | +- Hardware bugs such as Rowhammer, Meltdown or Spectre are analyzed |
| 120 | +- User error through analysis of how cryptography is used (usable security) |
| 121 | + |
| 122 | +- => Its odd when QKD considers itself to be outside of cryptography; the field is quite comprehensive |
| 123 | + |
| 124 | +## Open-source & Open-Science: Secure cryptography as a community process |
| 125 | + |
| 126 | +1. Cryptography is ultimately about creating trust |
| 127 | +2. Efficient, incremental peer review of implementations is essential |
| 128 | +3. Review of cryptographic proofs is essential |
| 129 | +4. Its not enough to say "I made this, this is secure" |
| 130 | + |
| 131 | +:::info |
| 132 | + |
| 133 | +@mullana Comic strip style illustration of cryptography as a community process |
| 134 | + |
| 135 | +Panel: |
| 136 | +- Single person saying: "Trust me, I built this, this is secure" |
| 137 | +- Other person saying: "Can I see it"; single person: "No, its a secret!"; "but what if you made a mistake" |
| 138 | +- Single person saying: "Just trust me already!" |
| 139 | +- (Narration: Its not enough for a single person to just build something) |
| 140 | + |
| 141 | +GAP |
| 142 | + |
| 143 | +- Cryptographer saying: "Don't trust me, please review my code. I might have made a mistake." |
| 144 | +- Big choir of cryptographers saying, single cryptographer: "Lets build it together to catch all our mistakes" |
| 145 | +- (Narration: All cryptographers make mistakes. The best ones are those who trust their work the least and who work together to build better code) |
| 146 | + |
| 147 | +::: |
| 148 | + |
| 149 | +## CAKD: Computational, Authenticated Key Distribution |
| 150 | + |
| 151 | +graphic representing TLS. |
| 152 | + |
| 153 | +Usually called a "key exchange". |
| 154 | + |
| 155 | +- Cheap |
| 156 | +- Fast |
| 157 | +- Secure |
| 158 | +- Extremely well analyzed |
| 159 | + |
| 160 | +## Key exchange – Security against quantum computers |
| 161 | + |
| 162 | +Graphic representing Rosenpass. |
| 163 | + |
| 164 | +- Most contemporary cryptography is not secure against quantum attacks |
| 165 | +- Migration is possible, Rosenpass is an example |
| 166 | +- Modest increase in resource usage |
| 167 | + |
| 168 | +## Security properties: Active vs passive |
| 169 | + |
| 170 | +:::info |
| 171 | + |
| 172 | +@mullana Two strip-style graphics to represent passive and active security. |
| 173 | + |
| 174 | +Passive – eavesdropping: Graphic of alice and bob communicating; attacker trying to listen in |
| 175 | + |
| 176 | +Active – man in the middle: Graphic of alice communicating with attacker; attacker communicating with bob |
| 177 | + |
| 178 | +::: |
| 179 | + |
| 180 | +## Security properties: Secrecy & Authenticity |
| 181 | + |
| 182 | +:::info |
| 183 | + |
| 184 | +@mullana Two strip-style graphics |
| 185 | + |
| 186 | +Secrecy: Panel of attacker trying to pry into a delivery package. |
| 187 | + |
| 188 | +Authenticity: Panel of attacker glueing a patch over the package address, but package contains a certificate when bob takes it out. |
| 189 | + |
| 190 | +::: |
| 191 | + |
| 192 | +## Security properties: Identity hiding, deniability |
| 193 | + |
| 194 | +:::info |
| 195 | + |
| 196 | +@mullana Three strip-style graphics |
| 197 | + |
| 198 | +Identity hiding: Panels of klandestine package-dropoff, delivery. Panel of empty address field. Panel of secret service delivery man being asked who is communicating: "I could tell you but I would have to kill you". |
| 199 | + |
| 200 | +Deniability: Panel of Bob going to a judge: "Alice sent me this". Panel of judge looking at package (with a spyglass?) "I can see no fingerprings. Guess you can't prove it". |
| 201 | + |
| 202 | +Non-repudiation: Panel of Bob going to a judge: "Alice sent me this". Panel of judge looking at a certificate: "This is clearly Alice's seal!" |
| 203 | + |
| 204 | +::: |
| 205 | + |
| 206 | +## Security properties: Forward secrecy |
| 207 | + |
| 208 | +:::info |
| 209 | + |
| 210 | +@mullana Three strip-style graphics |
| 211 | + |
| 212 | +Forward secrecy: Panel of package being received; panel of it being destroyed; panel of burglers not finding the package |
| 213 | + |
| 214 | +Forward secrecy provides no security against active attacks: Panel of burglars stealing the package during delivery. |
| 215 | + |
| 216 | +Forward secrecy can be broken when a cryptographic scheme itself is broken: Picture of burglars analyzing the remains of the destroyed package in a lab. |
| 217 | + |
| 218 | +::: |
| 219 | + |
| 220 | +## Everlasting secrecy: QKD improves on forward secrecy |
| 221 | + |
| 222 | +:::info |
| 223 | + |
| 224 | +@mullana One strip style illustration; one comic-style scientific illustration. |
| 225 | + |
| 226 | +Everlasting secrecy: Picture of destroying the package using some quantum device. Picture of attacker dissapointed at the remains being quantum. "Damn, these are quantum ashes." |
| 227 | + |
| 228 | +Comic style scientific illustration: |
| 229 | +A computer network of machines (feel free to represent as dots or circles), |
| 230 | +alice & bob communicating through multiple nodes on that network. |
| 231 | +The path through that network is marked one color; annotated "Software encryption". |
| 232 | +The path through that network is also marked in another color, but this path is interrupted at each node. Labeled "QKD". |
| 233 | +One of the nodes contains a graphic representing the attacker. |
| 234 | + |
| 235 | +Subtitle: Software encryption is end-to-end, but QKD is not. What if an attacker manages to take over a node. |
| 236 | + |
| 237 | +::: |
| 238 | + |
| 239 | +- We can not get end-to-end everlasting secrecy, so the forward secrecy that software provides probably still does more for practical security. |
| 240 | + |
| 241 | +## Limitations of QKD |
| 242 | + |
| 243 | +- Usually not open-source |
| 244 | +- Usually not open-hardware |
| 245 | +- Usually not peer-reviewed |
| 246 | +- Expensive |
| 247 | +- Inefficient |
| 248 | + |
| 249 | +| Security property | QKD | Software encryption | |
| 250 | +| --- | --- | --- | |
| 251 | +| Post-Quantum | check Green | Possibly Green | |
| 252 | +| Attacker-mode | passive RED | Active Green |
| 253 | +| Forward-secrecy | Pairwise Yellow | Green | |
| 254 | +| Everlasting-Secrecy | Pairwise Yellow | No Red | |
| 255 | +| Authenticity | cross Red | check Green | |
| 256 | +| Deniability | cross Red | check Green | |
| 257 | +| Non-repudiation | cross Red | check Green | |
| 258 | +| Identity hiding | cross Red | check Green | |
| 259 | + |
| 260 | +## QKD as a measure of hardware security |
| 261 | + |
| 262 | +:::info |
| 263 | + |
| 264 | +@mullana Once comic-strip style illustration |
| 265 | + |
| 266 | +Strip: Physical data cable connecting alice and bob across multiple nodes (computers). |
| 267 | +The entire cable and all nodes are protected by a wall and some guards. At some point the wall ends; |
| 268 | +instead each node – computer – now connects to an adjacent QKD device. The now QKD-protected cable (maybe with a glow |
| 269 | +in the previous qkd color) now exits the wall and enters another wall where it is terminated by a qkd device. |
| 270 | +Panel with engineer: "We managed to cut some costs by switching to QKD devices instead of guarding the entire length of the cable". |
| 271 | + |
| 272 | +::: |
| 273 | + |
| 274 | +## Hybrid QKD & Cryptography |
| 275 | + |
| 276 | +:::info |
| 277 | + |
| 278 | +@mullana Once comic-strip style illustration |
| 279 | + |
| 280 | +Graphic/strip: The aforementioned cable being attacked by burglars; they overwhelm the guards and hack the devices. |
| 281 | +Panel with engineer: "Our cable was attacked! Good thing we also used cryptography to protect the data." |
| 282 | + |
| 283 | +::: |
| 284 | + |
| 285 | +| Security property | QKD | Software encryption | Hybrid | |
| 286 | +| --- | --- | --- | |
| 287 | +| Post-Quantum | check Green | Supported Green | Supposed Green | |
| 288 | +| Attacker-mode | passive RED | Active Green | Active Green | |
| 289 | +| Forward-secrecy | Pairwise Yellow | Green | check green |
| 290 | +| Everlasting-Secrecy | Pairwise Yellow | No Red | Pairwise Yellow | |
| 291 | +| Authenticity | cross Red | check Green | check green | |
| 292 | +| Deniability | cross Red | check Green | check green | |
| 293 | +| Non-repudiation | cross Red | check Green | check Green | |
| 294 | +| Identity hiding | cross Red | check Green | check Green | |
| 295 | + |
| 296 | + |
| 297 | +- Expensive |
| 298 | +- Inefficient |
| 299 | +- Even if QKD devices are not well-reviewed, no security is lost by using them. |
0 commit comments