Igor Bjelakovic

Broadcast Capacity Region of Two-Phase Bidirectional Relaying

In a three-node network bidirectional communication between two nodes can be enabled by a half-duplex relay node with a decode-and-forward protocol. In the first phase, the messages of two nodes are transmitted to the relay node. In the second phase a re-encoded composition is broadcasted by the relay node. In this work the capacity region of the broadcast phase in terms of the maximal probability of error is determined. It is characterized by the mutual informations of the separate channels coupled by the common input.

Secrecy results for compound wiretap channels

We derive a lower bound on the secrecy capacity of a compound wiretap channel with channel state information at the transmitter which matches the general upper bound on the secrecy capacity of general compound wiretap channels given by Liang et al. [1], thus establishing a full coding theorem in this case. We achieve this with a stronger secrecy criterion and the maximum error probability criterion, and with a decoder that is robust against the effect of randomization in the encoding. This relieves us from the need of decoding the randomization parameter, which is in general impossible within this model. Moreover, we prove a lower bound on the secrecy capacity of a compound wiretap channel without channel state information and derive a multiletter expression for the capacity in this communication scenario.

Capacity of Gaussian MIMO bidirectional broadcast channels

We consider the broadcast phase of a three-node network, where a relay node establishes a bidirectional communication between two nodes using a spectrally efficient two-phase decode-and-forward protocol. In the first phase the two nodes transmit their messages to the relay node. Then the relay node decodes the messages and broadcasts a re-encoded composition of them in the second phase. We consider Gaussian MIMO channels and determine the capacity region for the second phase which we call the Gaussian MIMO bidirectional broadcast channel.

Capacity results for arbitrarily varying wiretap channels

In this work the arbitrarily varying wiretap channel AVWC is studied. We derive a lower bound on the random code secrecy capacity for the average error criterion and the strong secrecy criterion in the case of a best channel to the eavesdropper by using Ahlswedes robustification technique for ordinary AVCs. We show that in the case of a non-symmetrisable channel to the legitimate receiver the deterministic code secrecy capacity equals the random code secrecy capacity, a result similar to Ahlswedes dichotomy result for ordinary AVCs. Using this we can derive that the lower bound is also valid for the deterministic code capacity of the AVWC. The proof of the dichotomy result is based on the elimination technique introduced by Ahlswede for ordinary AVCs. We further prove upper bounds on the deterministic code secrecy capacity in the general case, which results in a multi-letter expression for the secrecy capacity in the case of a best channel to the eavesdropper. Using techniques of Ahlswede, developed to guarantee the validity of a reliability criterion, the main contribution of this work is to integrate the strong secrecy criterion into these techniques.

Achievable Rate Region of a Two Phase Bidirectional Relay Channel

In this work, the capacity region of the broadcast channel in a two phase bidirectional relay communication scenario is proved. Thereby, each receiving node has perfect knowledge about the message intended for the other node. The capacity region can be achieved using an auxiliary random variable taking two values, i.e., by the principle of time-sharing. The resulting achievable rate region of the two-phase bidirectional relaying includes the region which can be achieved by network coding applying XOR on the decoded messages at the relay node.

A Quantum Version of Sanov's Theorem

We present a quantum version of Sanovs theorem focussing on a hypothesis testing aspect of the theorem: There exists a sequence of typical subspaces for a given set Ψ of stationary quantum product states asymptotically separating them from another fixed stationary product state. Analogously to the classical case, the separating rate on a logarithmic scale is equal to the infimum of the quantum relative entropy with respect to the quantum reference state over the set Ψ. While in the classical case the separating subsets can be chosen universally, in the sense that they depend only on the chosen set of i.i.d. processes, in the quantum case the choice of the separating subspaces depends additionally on the reference state.

Classical Capacities of Compound and Averaged Quantum Channels

We determine the capacity of compound classical-quantum channels. As a consequence, we obtain the capacity formula for the averaged classical-quantum channels. The capacity result for compound channels demonstrates, as in the classical setting, the existence of reliable universal classical-quantum codes in scenarios where the only a priori information about the channel used for the transmission of information is that it belongs to a given set of memoryless classical-quantum channels. Our approach is based on a universal classical approximation of the quantum relative entropy which in turn relies on a universal hypothesis testing result.

Quantum capacity under adversarial quantum noise: arbitrarily varying quantum channels

AbstractWe investigate entanglement transmission over an unknown channel in the presence of a third party (called the adversary), which is enabled to choose the channel from a given set of memoryless but non-stationary channels without informing the legitimate sender and receiver about the particular choice that he made. This channel model is called an arbitrarily varying quantum channel (AVQC). We derive a quantum version of Ahlswede’s dichotomy for classical arbitrarily varying channels. This includes a regularized formula for the common randomness-assisted capacity for entanglement transmission of an AVQC. Quite surprisingly and in contrast to the classical analog of the problem involving the maximal and average error probability, we find that the capacity for entanglement transmission of an AVQC always equals its strong subspace transmission capacity. These results are accompanied by different notions of symmetrizability (zero-capacity conditions) as well as by conditions for an AVQC to have a capacity described by a single-letter formula. In the final part of the paper the capacity of the erasure-AVQC is computed and some light shed on the connection between AVQCs and zero-error capacities. Additionally, we show by entirely elementary and operational arguments motivated by the theory of AVQCs that the quantum, classical, and entanglement-assisted zero-error capacities of quantum channels are generically zero and are discontinuous at every positivity point.

Arbitrarily varying and compound classical-quantum channels and a note on quantum zero-error capacities

We consider compound as well as arbitrarily varying classical-quantum channel models. For classical-quantum compound channels, we give an elementary proof of the direct part of the coding theorem. A weak converse under average error criterion to this statement is also established. We use this result together with the robustification and elimination technique developed by Ahlswede in order to give an alternative proof of the direct part of the coding theorem for a finite classical-quantum arbitrarily varying channels with the criterion of success being average error probability. Moreover we provide a proof of the strong converse to the random coding capacity in this setting. The notion of symmetrizability for the maximal error probability is defined and it is shown to be both necessary and sufficient for the capacity for message transmission with maximal error probability criterion to equal zero. Finally, it is shown that the connection between zero-error capacity and certain arbitrarily varying channels is, just like in the case of quantum channels, only partially valid for classical-quantum channels.

Quantum capacity of a class of compound channels

In this paper we address the issue of universal or robust communication over quantum channels. Specifically, we consider a memoryless communication scenario with channel uncertainty which is an analog of the compound channel in classical information theory. We determine the quantum capacity of finite compound channels and arbitrary compound channels with an informed decoder. Our approach in the finite case is based on the observation that perfect channel knowledge at the decoder does not increase the capacity of finite quantum compound channels. As a consequence, we obtain a coding theorem for finite quantum averaged channels, the simplest class of channels with long-term memory. The extension of these results to quantum compound channels with uninformed encoder and decoder and infinitely many constituents remains an open problem.