No abstract

No abstract

We report results from a measurement study on the role of the most popular end-to-end security protocol Transport Layer Security (TLS) in the energy consumption of a mobile device. We measured energy consumed by TLS transactions between a Nokia N95 and several popular Web services over WLAN and 3G network interfaces. Our detailed analysis corroborates some earlier results but also reveals, contrary to earlier studies, that the transmission and I/O energy, both in the TLS handshake and the record protocol, far exceed the required computational energy by the actual cryptographic algorithms and that with transactions larger than 500KB, the energy required to transmit the actual data clearly outranks the TLS energy overhead. In addition, we note that the energy consumption varies remarkably between measured services. I. INTRODUCTIONIn the last few years, mobile devices have evolved significantly in terms of power, throughput, and in terms of new functionalities, but they are still severely constrained by limited battery life-time. Secure communications are achieved by employing security protocols, which are based on cryptographic algorithms.Executing certain cryptographic algorithms requires rather intensive computations. Therefore, their energy consumption on these battery powered devices is naturally a concern. In this paper, we study the energy consumption of Transport Layer Security (TLS), which is used to establish a secure communication channel between two end hosts and exchange data over that channel. TLS is a transport level protocol that uses asymmetric and symmetric encryption algorithms and hash algorithms in order to provide data secrecy, authentication of the communication parties, and data integrity for applications in a transparent manner.We use a Symbian mobile device (Nokia N95) to establish TLS connections to different web services, such as electronic email or social networks, over both WLAN and 3G network interfaces and measure the energy consumption. TLS comprises two phases. First, a security association is created through a handshake protocol after which the actual data is transferred over an encrypted and integrity protected channel. We perform detailed analysis of the different steps involved in TLS transactions, compute the amount of energy overhead in a TLS transaction, and explain the major causes that determine this overhead.Energy consumption of different cryptographic algorithms as well as the performance of TLS on PDAs and in computers

Abstract-This paper presents results from performance measurements carried out in a Differentiated Services WAN connecting three major cities in Southern Finland. The target of the measurements was to collect precise information about the level of service, experienced by a user transmitting files with FTP in a network that implements service differentiation by using drop precedence levels and separate classes for realtime and non-realtime traffic. Special attention was paid to the behaviour of competing TCP flows produced by users with different service level specifications. Background traffic from ISP's network was injected into our test network in order to create traffic profiles which are typical in a core network. The results show clearly that the SIMA model used for the implementation of the DS mechanisms in the network nodes can provide service differentiation in a useful and fair manner in traffic conditions varying from light load to severe overload.

No abstract