This paper considers relay assisted transmission for multiuser networks when the relay has no access to the codebooks used by the transmitters. The relay is called oblivious for this reason. Of particular interest is the generalized compress-and-forward (GCF) strategy, where the destinations jointly decode the compression indices and the transmitted messages, and their optimality in this setting. The relay-to-destination links are assumed to be out-of-band with finite capacity. Two models are investigated: 1) the multiple access relay channel (MARC) and 2) the interference relay channel (IFRC). For the MARC with an oblivious relay, a new outerbound is derived and it is shown to be tight by means of achievability of the capacity region using GCF scheme. For the IFRC with an oblivious relay, a new strong interference condition is established, under which the capacity region is found by deriving a new outerbound and showing that it is achievable using GCF scheme. The result is further extended to establish the capacity region of M-user MARC with an oblivious relay, and multicast networks containing M sources and K destinations with an oblivious relay.

This paper investigates the degrees of freedom (DoF) of the L-cluster, K-user MIMO multiway relay channel, where users in each cluster wish to exchange messages within the cluster, and they can only communicate through the relay. A novel DoF upper bound is derived by providing users with carefully designed genie information. Achievable DoF is identified using signal space alignment and multiple-access transmission. For the two-cluster MIMO multiway relay channel with two users in each cluster, the DoF is established for the general case when users and the relay have arbitrary number of antennas, and it is shown that the DoF upper bound can be achieved using signal space alignment or multiple-access transmission, or a combination of both. The result is then generalized to the three user case. For the L-cluster K-user MIMO multiway relay channel in the symmetric setting, conditions under which the DoF upper bound can be achieved are established. In addition to being shown to be tight in a variety of scenarios of interests of the multiway relay channel, the newly derived upperbound also establishes the optimality of several previously established achievable DoF results for multiuser relay channels that are special cases of the multiway relay channel.

This paper investigates the degrees of freedom (DoF) of the two-cluster multi-way relay channel. Two cases are investigated: the case when there are 2 users in each cluster with arbitrary number of antennas and the case when there are 3 users in each cluster in the symmetric setting, i.e., all the users have the same number of antennas. For the 2-user case, a DoF upper bound is derived based on cut set bound by allowing user cooperation between clusters. Conditions when the DoF upper bound can be achieved using signal space alignment are established based on the relative number of antennas between the users and the relay. For the 3-user case, a new DoF upper bound is derived using genie-aided approach and channel enhancement. The DoF upper bound can be achieved using signal space alignment for several scenarios of interests. The results point out the insight that increasing the number of users in each cluster cannot provide a further DoF gain compared to the 2-user case, when the relay has limited number of antennas.

This paper investigates the degrees of freedom (DoF) of the L-cluster LT-user MIMO multi-way relay channel. A DoF upperbound is derived by providing users with carefully designed genie information, and properly enhancing the received signal of one of the users. For the L-cluster LT-user MIMO multi-way relay channel in the symmetric setting, conditions under which the DoF upperbound can be achieved using either multiple access transmission or signal space alignment are established, demonstrating that the newly derived upperbound is the first tight DoF upperbound for the general MIMO multi-way relay channel. Additionally, this new upperbound proves the optimality of the achievable DoF for several special cases of the MIMO multi-way relay channel obtained in previous works. The results provide the insight that, with fixed spatial dimension at the relay, increasing the number of users and clusters cannot provide any DoF gain. In addition, it is observed that allowing three or more users to share the spatial dimension of the relay cannot provide any DoF gain.

Channel state information (CSI) at the transmitters (CSIT) is of importance for interference alignment schemes to achieve the optimal degrees of freedom (DoF) for wireless networks. This paper investigates the impact of half-duplex relays on the DoF of the X channel and the interference channel when the transmitters are blind in the sense that no CSIT is available. In particular, it is shown that adding relay nodes with global CSI to the communication model is sufficient to recover the DoF that is the optimal for these models with global CSI at the transmitters. The relay nodes in essence help steer the directions of the transmitted signals to facilitate interference alignment to achieve the optimal DoF with CSIT. The general M × N X channel with relays and the -user interference channel are both investigated, and sufficient conditions on the number of antennas at the relays and the number of relays needed to achieve the optimal DoF with CSIT are established. Using relays, the optimal DoF can be achieved in finite channel uses. The DoF for the case when relays only have delayed CSI is also investigated, and it is shown that with delayed CSI at the relay the optimal DoF with full CSIT cannot be achieved. Special cases of the X channel and interference channel are investigated to obtain further design insights.

This paper investigates the impact of half-duplex relays on the degrees of freedom (DoF) of the K-user X channel when the transmitters are blind in the sense that no channel state information (CSI) at the transmitter (CSIT) is available. It is shown that adding relay nodes with CSI to the communication model is sufficient to recover the optimal DoF with CSIT. In particular, the relay nodes help steer the directions of the transmitted signals to facilitate interference alignment to achieve optimal DoF. It is shown that one relay node with K-1 antennas is sufficient to achieve the optimal DoF. When relays do not have multiple antennas, it is sufficient to have (K - 1)2relays to achieve the optimal DoF. The achievability is established using partial interference alignment and joint beamforming. DoF for the case when relays only have delayed CSI is also investigated.

In this work, we investigate the impact of placing a half-duplex relay in the X channel in the presence of limited channel state information (CSI). Specifically, we consider the scenarios when the sources have no CSI, while the relay either has perfect CSI or delayed CSI. We develop transmission strategies for each scenario to facilitate interference alignment with the help of the relay. We show that for both scenarios, with single antenna at each source and destination, degree of freedom 4/3 can be achieved. This matches the degree of freedom established in previous work when perfect CSI is available at all nodes. For the case with perfect CSI at the relay, relay only needs to have 1 antenna, and can perform beamforming with the sources to facilitate interference alignment using the proposed scheme. For the case with delayed CSI at the relay, the relay can compensate the staleness of the CSI by adding an antenna. An important observation from this work is that relay can increase the degrees of freedom for the X channel when the sources have no CSI, in contrast with the case when perfect CSI is available at all nodes, where a relay cannot improve the degrees of freedom. The results are also extended to multi-antenna settings.

This paper studies the Gaussian interference channel (IC) with a relay, which transmits and receives in a band that is orthogonal to the IC. The channel associated with the relay is thus an out-of-band relay channel (OBRC). The focus is on a symmetric channel model, in order to assess the fundamental impact of the OBRC on the signal interaction of the IC, in the simplest possible setting. First, the linear deterministic model is investigated and the sum capacity of this channel is established for all possible channel parameters. In particular, it is observed that the impact of OBRC, as its links get stronger, is similar to that of output feedback for the IC. The insights obtained from the deterministic model are then used to design achievable schemes for the Gaussian model. The interference links are classified as extremely strong, very strong, strong, moderate, weak, and very weak. For strong and moderate interference, separate encoding is near optimal. For very strong and extremely strong interference, the interference links provide side information to the destinations, which can help the transmission through the OBRC. For weak or very weak interference, an extension of the Han-Kobayashi scheme for the IC is utilized, where the messages are split into common and private. To achieve higher rates, it is beneficial to further split the common message into two parts, and the OBRC plays an important role in decoding the common message. It is shown that our strategy achieves the symmetric capacity to within 1.14625 bits per channel use with duplexing factor 0.5, and 1.27125 bits per channel use for arbitrary duplexing factors, for all channel parameters. An important observation from the constant gap result is that strong interference can be beneficial with the presence of an OBR.

This work studies the Gaussian interference channel (IC) with an out-of-band relay (OBR) from an information-theoretical perspective. The relay works in bands orthogonal to the IC. The focus is on a symmetric channel model, in order to understand the fundamental impact of the OBR on the signal interaction of the IC, in the simplest possible setting. We classify the interference links as extremely strong, very strong, strong, moderate, weak, and very weak. For strong and moderate interference, separate encoding is close to optimal. For very strong and extremely strong interference, the interference links provide side information to the destinations, which can help the transmission through the OBRC. In particular, when interference is extremely strong, the channel acts as if there are two disjoint OBRC helping each source-destination pair. For moderate or weak interference, we use the Han-Kobayashi scheme for the IC, where we split the messages into common and private messages. We find that it is beneficial to further split the common message into two parts, and the OBRC plays an important role in decoding the common messages. It is shown that our strategy achieves the symmetric capacity to within 1.14625 bits per channel use for all channel parameters.

We consider the Gaussian interference channel with an intermediate relay as a main building block for cooperative interference networks. On the achievability side, we consider compress-and-forward based strategies. Specifically, a generalized compress-and-forward strategy, where the destinations jointly decode the compression indices and the source messages, is shown to improve upon the compress-and-forward strategy which sequentially decodes the compression indices and source messages, and the recently proposed generalized hash-and-forward strategy. We also construct a nested lattice code based compute-and-forward relaying scheme, which outperforms other relaying schemes when the direct link is weak. In this case, it is shown that, with a relay, the interference link can be useful for decoding the source messages. Noting the need for upperbounding the capacity for this channel, we propose a new technique with which the sum rate can be bounded. In particular, the sum capacity is upperbounded by considering the channel when the relay node has abundant power and is named potent for that reason. For the Gaussian interference relay channel with potent relay, we study the strong and the weak interference regimes and establish the sum capacity, which, in turn, serve as upperbounds for the sum capacity of the GIFRC with finite relay power. Numerical results demonstrate that upperbounds are tighter than the cut-set bound, and coincide with known achievable sum rates for many scenarios of interest. Additionally, the degrees of freedom of the GIFRC are shown to be 2 when the relay has large power, achievable using compress-and-forward.

We consider the Gaussian interference channel with an intermediate relay. The known achievable schemes for this channel are based on decode-and-forward (DF) relaying, whose performance depends on the signal-to-noise (SNR) ratio of the received signal at the relay. Consequently, when the source-to-relay (S-R) links are weak, the resulting achievable rates have room for improvement. In this work, we design achievable schemes that provide this improvement. First, we consider compress-and-forward (CF) relaying to overcome the weakness of the DF schemes. Second, we consider employing structured codes and observe that when the signal strength from the direct link is subject to severe attenuation, using nested lattice codes yields higher rates than both the DF and CF schemes. Numerical results are presented to demonstrate the performance of different achievable schemes with the Potent Relay Outerbound derived in our previous work.

We consider the ergodic fading Gaussian interference relay channel (EF-GIFRC) with individual power constraints at the nodes. Aiming at design insights to emerge from the optimal power allocation, we focus on scenarios similar to that of the degraded/reversely degraded relay channels. In particular, we focus on models where the source-to-relay (S - R) links are either stronger than direct links, or completely blocked, i.e., with an “on-and-off” relay. To characterize the capacity of EF-GIFRC with an on-and-off relay, we first investigate the parallel interference relay channel. We propose an achievable scheme based on partial decode-and-forward (DF) strategy and show that the capacity for the parallel IFRC can be achieved under strong interference and degradedness conditions. Based on the achievable rate region for parallel IFRC, we propose an achievable rate region for EF-GIFRC, and present the properties of optimal power allocation. We also present a sum capacity result when the EF-GIFRC satisfies certain channel conditions.

We consider the Gaussian interference channel with an intermediate relay. The relay is assumed to have abundant power and is named potent for that reason. A main reason to consider this model is to find good outerbounds for the Gaussian interference relay channel (GIFRC) with finite relay power. By setting the power of the relay constraint to infinity, we show that the capacity region is asymptotically equivalent to the case when the relay-destination links are noiseless and orthogonal to other links. The capacity region of the latter provides an outerbound for the GIFRC with finite relay power. We then show the capacity region of the former can be upper bounded by a single-input-multiple-output interference channel with an antenna common to both receivers. To establish the sum capacity of this channel, we study the strong and the weak interference regimes. For both regimes, we show that the upperbounds we find are achievable, thus establishing the sum capacity of GIFRC with the potent relay. Both results, in turn, serve as upperbounds for the sum capacity of the GIFRC with finite relay power. Numerical results show that the upperbounds are close to the known achievable rates for many scenarios of interest.