We calculate the semiclassical magnetoresistivity r xx ͑B͒ of noninteracting fermions in two dimensions moving in a weak and smoothly varying random potential or random magnetic field. We demonstrate that in a broad range of magnetic fields the non-Markovian character of the transport leads to a strong positive magnetoresistance. The effect is especially pronounced in the case of a random magnetic field where r xx ͑B͒ becomes parametrically much larger than its B 0 value. 73.50.Jt The magnetoresistance (MR) is one of the most frequently studied characteristics of the two-dimensional electron gas (2DEG). When the effect of disorder is described by a collision integral within the semiclassical Boltzmann equation approach, the resistivity tensorr͑B͒ for an isotropic system has the Drude formwhere n is the carrier density, m the effective mass, v c eB͞mc the cyclotron frequency, and t the transport scattering time. In particular, the longitudinal resistivity r xx is independent of the magnetic field B, r xx ͑B͒ r 0 ϵ m͞e 2 nt, irrespective of the form of the impurity collision integral. This result is solely determined by the Markovian character of the transport assumed in the Boltzmann equation description. Deviations from the constant r xx ͑B͒ are conventionally termed a positive͞negative MR, depending on the sign of the deviation. The negative MR [1] induced by the suppression of the quantum interference correction by the magnetic field is a famous manifestation of weak localization. Another source of negative MR is the Altshuler-Aronov correction to the conductivity due to enhancement of the electron-electron interaction by the diffusive motion of particles [1]. Both these effects are of quantum nature and lead to a correction of order e 2 ͞h to the conductivity s xx , and thus to a small correction to r xx .However, as we will show, already at the classical level there exists a nontrivial MR which can be much stronger than the quantum one, if the correlation length d of disorder is sufficiently large, k F d ¿ 1 (where k F is the Fermi wave vector). This is due to memory effects which are neglected in the collision integral description of disorder.Transport properties of the 2DEG in a smooth random potential (RP) V ͑r͒ are of particular interest, since in currently fabricated high-mobility semiconductor heterostructures the disorder has long-range character. The high mobility of these samples is achieved by placing the charged donor ions in a layer separated by a large distance d (k F d ϳ 10) from the 2DEG plane. Assuming the positions of these impurities to be statistically distributed with a sheet density n i , the correlation function W V ͑r 2 r 0 ͒ ͗V ͑r͒V ͑r 0 ͒͘ is given in momentum space bỹ W V ͑q͒ ͑ph 2 ͞m͒ 2 n i e 22qd .(2)A new type of transport problem occurs in these systems when a large magnetic field B Ӎ B 1͞2 2͑hc͞e͒n is applied such that the lowest Landau level is approximately half filled. The metallic state then observed has been described [2] in terms of composite fermions (CF's) moving in ...

We study the low-frequency behavior of the ac conductivity σ(ω) of a two-dimensional fermion gas subject to a smooth random potential (RP) or random magnetic field (RMF). We find a nonanalytic ∝ |ω| correction to Re σ, which corresponds to a 1/t 2 long-time tail in the velocity correlation function. This contribution is induced by return processes neglected in Boltzmann transport theory. The prefactor of this |ω|-term is positive and proportional to (d/l) 2 for RP, while it is of opposite sign and proportional to d/l in the weak RMF case, where l is the mean free path and d the disorder correlation length. This non-analytic correction also exists in the strong RMF regime, when the transport is of a percolating nature. The analytical results are supported and complemented by numerical simulations.

No abstract

The development of energy-efficient applications and protocols is one of the most important issues in Wireless Sensor Networks (WSN). However, most publications up to now avoid time consuming realistic energy evaluations and oversimplify their evaluation with regard to energy-efficiency. This work aims at lowering the barrier for realistic energy evaluations. We focus on a generic application that simply transmits one packet using TinyOS Low Power Listening (LPL), which we evaluate using the WSN testbed SANDbed. Our results disprove some intuitive expectations. For example, we show that transmitting packets with a large payload can be cheaper in terms of energy consumption than a small payload. As packet transmission is part of almost any WSN application, the results shown are important to many WSN protocol evaluations. As an addition, we contribute our lessons learned by discussing the most important challenges and pitfalls we faced during our evaluation.