Application of trusted computing to secure P2P networking

Abstract

Peer-to-peer (P2P) networking is based on a distributed application architecture that covers a diverse set of network types. In pure P2P overlays, every node in the network acts as a "servent", they act as a server and a client simultaneously. Highlighting the fact that there is a lack of any centralised control, acting all nodes as equals. This networks have become very popular in the form of file-sharing networks. It is based on the idea of sharing any type of resources between all nodes. Peers are both suppliers and consumers of resources. As nodes arrive, the total capacity of the system also increases. In contrast, in a traditional client-server architecture, clients share only their demands with the system, but not their resources. In this case, as more clients join the system, fewer resources are available to serve each client. The decentralized nature of P2P networks also increases robustness because if a part of the system fails, it will not stop the entire system from working. As mentioned in there is a need to provide robust access control, data integrity, confidentiality and accountability services. In order to prevent other nodes from impersonating or creating an arbitrary amount of bogus nodes, all distributed systems must have a unique, undeniable and verifiable identifier for each node. The foundation of stable and verifiable identities is required to build nodes with secure parameters. The use of unsecure nodes may allow remote access to files on a victim's computer or even compromise the entire network, this is explained in. Finding suitable measures against the increasing variety of software-based attacks is a di cult task. In particular, pseudospoo ng attacks, in which malicious parties claim multiple identities and disrupt the operation of P2P networks. The Trusted Computing paradigm offers a very useful and powerful set of security features to improve on computer systems. The Trusted Computing Group (TCG) specifies the Trusted Platform Module (TPM), wich allows to provide cryptographically qualified and tamper-resilient statements on the software configuration of a machine. As described in, the use of protected cryptographic mechanisms alone is not suffi cient to convince remote machines or human users that a complex software service can actually be trusted. To enable a decision based on the statements made by a TPM and the associated trust levels, keys need to be vouched for. This requires a Public Key Infrastucture (PKI). Allowing a remote and indipendant party to decide on the trustworthiness of a host or a particular service. The TPM provides authenticity and allows unique identification of a platform, therefore creating a privacy problem. To circumvent this problem the TCG proposed a trusted third party, the Privacy Certification Authority (PrivacyCa). A particular incarnation of the PKI concept is the PrivacyCa, a trusted service capable of reporting its state to clients. It confirms that keys are protected by a specification-compliant TPM implementation and thus may be trusted under certain conditions, but without revealing the specific identity of the TPM and the user. In this work we provide a solution for creating unique, undeniable and verifiable identifiers for P2P networks by using the security features that the TPM o ers. We implement a protocol called Trusted Authentication protocol (TAP) o ering two possible solutions for authentication. One based on machines with TPM hardware enabled and a PrivacyCa that dictates if a machine can be trusted. The other solution is based on the authentication of machines using their own TPM without a PrivacyCa. The main goals of a secure P2P would be to identify and avoid malicious nodes, this is achieved by secure authentication of machines via TPM. -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Las redes P2P están basadas en una arquitectura distribuida que cubre un conjunto de tipos de redes. La propiedad común compartida por casi todas las redes P2P es la ausencia de un control centralizado. Este tipo de redes son la antítesis del modelo cliente-servidor tradicional. Se han hecho muy populares para compartir archivos. En este proyecto se demuestra como algunas propiedades de la especificación TCG (Trusted Computing Group) pueden ser utilizadas para mejorar la seguridad en redes P2P. Los pasos realizados para elaborar el proyecto han sido: – En primer lugar se ha llevado a cabo un estudio del estado del arte en el cual se han investigado las distintas soluciones que ofrece Trusted Computing y su aplicación en redes P2P. – Aprendizaje y uso de las herramientas TPM Tools, conjunto de programas y librerías. – A continuación se ha realizado el diseño del protocolo basado en una publicación del grupo IAIK SCOS el cual explica una solución de forma teórica. – También se ha llevado a cabo una familiarización inicial de los algoritmos criptográficos y mecanismos como cifrado simétrico y asimétrico, generación de número aleatorios de forma segura, funciones resumen (hash) y firmas digitales. Utilizando estos mecanismos, se ha diseñado un protocolo seguro utilizando distintos métodos del área de Trusted Computing. Un protocolo utiliza el concepto de PrivacyCA para autenticación y protección de la privacidad mientras que el otro protocolo se basa únicamente en la utilización del módulo TPM (Trusted Platform Module). Este último diseño es el que más encaja con la filosofía P2P dado que la ausencia de la entidad PrivacyCA deja a todos los nodos de la red como iguales. De esta forma todos los nodos trabajan de forma distribuida.