1. Introduction -- 1.1 Insecure computers in a hostile world -- 1.2 A vision for a better world -- 1.3 Overview: building up from a firm foundation -- 1.4 Bootstrapping trust in a commodity computer -- 1.5 Securely executing code on a commodity computer -- 1.6 Leveraging secure code execution to improve network protocols -- 1.7 Secure code execution despite untrusted software and hardware -- 1.8 Summary of contributions --

2. Background and related work in trust establishment -- 2.1 What do we need to know? Techniques for recording platform state -- 2.1.1 Recording code identity -- 2.1.2 Recording dynamic properties -- 2.1.3 Which property is necessary? -- 2.2 Can we use platform information locally? -- 2.2.1 Secure boot -- 2.2.2 Storage access control based on code identity -- 2.3 Can we use platform information remotely? -- 2.3.1 Prerequisites -- 2.3.2 Conveying code measurement chains -- 2.3.3 Privacy concerns -- 2.4 How do we make sense of platform state? -- 2.4.1 Coping with information overload -- 2.4.2 Focusing on security-relevant code -- 2.4.3 Conveying higher-level information -- 2.5 Roots of trust -- 2.5.1 General-purpose tamper-resistant and tamper-responding devices -- 2.5.2 General-purpose devices without dedicated physical defenses -- 2.5.3 Special-purpose minimal devices -- 2.5.4 Research solutions without hardware support -- 2.5.5 Cryptographic protocols -- 2.6 Validating the process -- 2.7 Applications -- 2.7.1 Real world -- 2.7.2 Research proposals -- 2.8 Human factors and usability -- 2.8.1 Trustworthy verifier device -- 2.8.2 Using your brain to check a computer -- 2.8.3 Pairing two trustworthy devices -- 2.9 Limitations -- 2.9.1 Load-time vs. run-time guarantees -- 2.9.2 Hardware attacks -- 2.10 Additional reading -- 2.11 Summary --

3. Bootstrapping trust in a commodity computer -- 3.1 Problem definition -- 3.1.1 Informal problem description -- 3.1.2 Formal model -- 3.2 Potential solutions -- 3.2.1 Removing network access -- 3.2.2 Eliminating malware -- 3.2.3 Establishing a secure channel -- 3.3 Preferred solutions -- 3.4 Summary --

4. On-demand secure code execution on commodity computers -- 4.1 Problem definition -- 4.1.1 Adversary model -- 4.1.2 Goals -- 4.2 Flicker architecture -- 4.2.1 Flicker overview -- 4.2.2 Isolated execution -- 4.2.3 Multiple flicker sessions -- 4.2.4 Interaction with a remote party -- 4.3 Developer's perspective -- 4.3.1 Creating a PAL -- 4.3.2 Automation -- 4.4 Flicker applications -- 4.4.1 Stateless applications -- 4.4.2 Integrity-protected state -- 4.4.3 Secret and integrity-protected state -- 4.5 Performance evaluation -- 4.5.1 Experimental setup -- 4.5.2 Microbenchmarks -- 4.5.3 Stateless applications -- 4.5.4 Integrity-protected state -- 4.5.5 Secret and integrity-protected state -- 4.5.6 Impact on suspended operating system -- 4.5.7 Major performance problems -- 4.6 Architectural recommendations -- 4.6.1 Launching a PAL -- 4.6.2 Hardware memory isolation -- 4.6.3 Hardware context switch -- 4.6.4 Improved TPM support for flicker -- 4.6.5 PAL exit -- 4.6.6 PAL life cycle -- 4.6.7 Expected impact -- 4.6.8 Extensions -- 4.7 Summary --

5. Using trustworthy host-based information in the network -- 5.1 Problem definition -- 5.1.1 Architectural goals -- 5.1.2 Assumptions -- 5.2 The assayer architecture -- 5.2.1 Overview -- 5.2.2 Assayer components -- 5.2.3 Protocol details -- 5.2.4 User privacy and client revocation -- 5.3 Potential attacks -- 5.3.1 Exploited clients -- 5.3.2 Malicious clients -- 5.3.3 Rogue verifiers -- 5.3.4 Rogue filters -- 5.4 Case studies -- 5.4.1 Spam identification -- 5.4.2 Distributed denial-of-service (DDoS) mitigation -- 5.4.3 Super-spreader worm detection -- 5.5 Implementation -- 5.5.1 Client architecture -- 5.5.2 Client verification -- 5.5.3 Traffic annotation -- 5.5.4 Filter -- 5.6 Evaluation -- 5.6.1 Client verification -- 5.6.2 Client annotations -- 5.6.3 Filter throughput -- 5.6.4 Internet-scale simulation -- 5.7 Potential objections -- 5.7.1 Why not collect information on the local router? -- 5.7.2 Is this really deployable incrementally? -- 5.8 Summary --

6. Verifiable computing: secure code execution despite untrusted software and hardware -- 6.1 Overview -- 6.2 Cryptographic background -- 6.2.1 Yao's garbled circuit construction -- 6.2.2 The security of Yao's protocol -- 6.2.3 Fully homomorphic encryption -- 6.3 Problem definition -- 6.3.1 Basic requirements -- 6.3.2 Input and output privacy -- 6.3.3 Efficiency -- 6.4 An efficient verifiable-computation scheme with input and output privacy -- 6.4.1 Protocol definition -- 6.4.2 Proof of security -- 6.4.3 Proof of input and output privacy -- 6.4.4 Efficiency -- 6.5 How to handle cheating workers -- 6.6 Summary --

7. Conclusion -- Bibliography -- Author's biography.