SLIDE 1
Reliable M IX Cascade Networks t hrough Reput at ion Roger D - - PowerPoint PPT Presentation
Reliable M IX Cascade Networks t hrough Reput at ion Roger D - - PowerPoint PPT Presentation
Reliable M IX Cascade Networks t hrough Reput at ion Roger D ingledine, Reput at ion T echnologies Paul Syverson, Naval Research L ab 1 W ays of Improving Reliability B uild prot ocols wit h provable robust ness guarant ees Provide
SLIDE 2
SLIDE 3
W ays of Improving Reliability B uild prot ocols wit h provable robust ness guarant ees Provide economic incent ives for reliability Add reput at ion t o \ improve" reliability D ist inct ion between reliability and robust ness
3
SLIDE 4
Relat ed W ork M IXes ( Chaum) Robust M IX-net s ( Flash M ix, Universally Veri able M IX) D eployed Remailer Syst ems ( cypherpunks, M ixmast er) Remailer st at ist ics ( L evien’s st at ist ics, Jack B Nymble 2)
4
SLIDE 5
SLIDE 6
T hreat M odel | Adversary can: Passively read all t ra c Compromise some fract ion of t he M IXes ( Insert , modify, delay, or drop messages)
6
SLIDE 7
Previous paper at Info Hiding 4 M IXes writ e per-hop receipt s t o prove good service; wit nesses verify and t ally failure claims. B ut : Global wit nesses are t rust and communicat ion bot t lenecks O wning high reput at ion nodes means you own more pat hs?
7
SLIDE 8
W hat ’s a M IX cascade? Fixed pat h t hrough t he M IX network L onger cascades ) lower chance all bad nodes ) more anonymity L onger cascades ) lower chance all good nodes ) less reliability Cascades provide more defense against int ersect ion at t ack.
8
SLIDE 9
D esign O verview Cascades rearrange periodically ( e.g., daily) A node fails it s own cascade if it det ect s misbehavior Nodes send t est messages t o monit or t heir cascades Senders can demonst rat e decrypt ions t o show failure
SLIDE 10
Communal Randomness Goal: collaborat ing nodes cannot predict t he cascades Cent ralized ( but veri able) for convenience All nodes commit , t hen all reveal B ut nodes can in uence communal value by not revealing?
SLIDE 11
Heurist ics for picking cascades Increase cost of breaking anonymity
SLIDE 12
At t ack: Creepe589a/ F15 ( D eat h) ]T J/ F219.833 T f 0 -184.262 -72.727d[( A)
SLIDE 13
Need t o limit number of bad nodes in network Proof of work, proof of bandwidt h not st rong enough Advogat o t rust met ric: Number of bad nodes cert i ed is based on number of confused nodes ( good nodes t hat might cert ify bad nodes) Cert ify by t rustwort hiness, not expect ed performance
13
SLIDE 14
So how do we choose cascades? Pick a t arget safety fact or S ( eg 1 in 105 pat hs bad) Choose rst cascade randomly from large enough pool of high-reput at ion nodes Replace chosen nodes t o maint ain pool size W hen pool cont ains all remaining nodes, just build remaining cascades randomly
14
SLIDE 15
SLIDE 16
D et ect ing M isbehavior Ent ry point : Incoming messages reject ed? Inside cascade: M essages replaced wit h dummy messages? Exit point : M essages not delivered?
16
SLIDE 17
D et ect ing M isbehavior at Ent ry Point Alice can send int o any node. T hey all deliver t o t 0e head. T hus nodes can insert indist inguishable t est messages Alice get s a receipt ( if not , s0e t ries elsewhere) Head publishes bat ch snapshot ( hashes of messages) If message not in snapshot , receipt proves misbehavior
17
SLIDE 18
SLIDE 19
D et ect ing M isbehavior at Exit Point T ail bounces t ra c t o all nodes. All nodes deliver. If insert ed t est message doesn’t arrive, somebody failed. O pt imize: if t ail collect s a delivery receipt , no broadcast .
19
SLIDE 20
T est messages Nodes reuse recipient addresses in t est messages Reusing addresses helps prot ect against t ime-based int ersect ion at t ack
20
SLIDE 21
Q uality of Service, Resource M anagement Nodes send failure messages and hourly heart beat s t o Reput at ion Servers Users compare advert ised Q oS and reput at ion from each
SLIDE 22