A fog radio access network (F-RAN) is studied, in which KTedge nodes (ENs) connected to a cloud server via orthogonal fronthaul links, serve KRusers through a wireless Gaussian interference channel. Both the ENs and the users have finite-capacity cache memories, which are filled before the user demands are revealed. While a centralized placement phase is used for the ENs, which model static base stations, a decentralized placement is leveraged for the mobile users. An achievable transmission scheme is presented, which employs a combination of interference alignment, zero-forcing and interference cancellation techniques in the delivery phase, and the \textit{normalized delivery time} (NDT), which captures the worst-case latency, is analyzed.

A KT× KRcache-aided wireless interference network, in which both the transmitters and the receivers are equipped with cache memories is studied. Each user requests one file from a library of N popular files. The goal is to design the cache contents without the knowledge of the particular user demands, such that all possible demand combinations can be satisfied reliably over the interference channel. The achievable sum degrees-of-freedom (sDoF) and the normalized delivery time (NDT) are studied for centralized and decentralized network architectures, respectively. First, using a combination of interference alignment (IA), zero-forcing (ZF) and interference cancellation (IC) techniques, a novel caching and transmission scheme for centralized networks is introduced, and it is shown to improve the sDoF upon the state-of-the-art. Then, the NDT is studied when the content placement at the receiver caches is carried out in a decentralized manner. Our results indicate that, for this particular network architecture, caches located at the receiver side are more effective than those at the transmitter side in order to reduce the NDT.

Fundamental limits of communication is studied in a 4 × 4 interference network, in which the transmitters are equipped with cache memories. Each of the receivers requests one file from a library of N equal-size files. The caches at the transmitters are filled without the knowledge of the user demands, such that all possible demand combinations can be satisfied reliably over the interference channel. The achievable normalized delivery time (NDT) is studied under centralized cache placement. By combining the interference alignment (IA) and zero-forcing (ZF) techniques, a novel caching and transmission scheme is presented, and is shown to be optimal for all possible cache sizes; fully characterizing the NDT for the 4× 4 interference network with caches at the transmitter side.

Tone reservation (TR) is a well-known and practical strategy to reduce peak-to-average power ratio (PAPR) in OFDM signals. In TR some unused or reserved subcarriers are used to accommodate a peak-cancelling signal that lowers the PAPR of the time domain OFDM signal. Several methods have been proposed to optimise the peak-cancelling signal, amongst which the so called active-set method has been shown to converge very quickly to a minimum PAPR solution for a real baseband signal with no additional power constraints. However, for a general complex baseband signal and/or when the peak-cancelling signal has to be power limited, there is no guarantee of optimality. Moreover, in the complex baseband model, the TR optimisation problem is a quadratically constrained quadratic program (QCQP), which has a prohibitive computational cost. In this paper we introduce a tree-search approach to designing the TR signal based on the sphere decoding/encoding principle. The method is proved to provide the minimum PAPR solution in the limit of fine lattice quantisation. Simulation results show a significant margin of improvement over the best known active-set method.

While obtaining the channel state information (CSI) required to perform Interference Alignment (IA) in a centralised MIMO network is a challenge, this is compounded in unplanned wireless networks that may be decentralised - not to mention composed of devices with varying degrees of complexity and capability. This is of particular interest to heterogeneous networks and the Internet of Things (IoT). The disparity in antenna number caused by this variation in capability can be taken advantage of, increasing the sum rate of the network providing the users are aware of the capabilities of their neighbours. This paper presents a method for estimating with a high degree of certainty the number of antennas each user possesses, endowing the receiver with the vital CSI required before channel equalisation or estimation can take place. The method is correlation based and so is not restricted to full rank channels where the receiving device has an equal or greater number of antennas than the transmitter. This permits the use of the method in situations where the rank of the channel is restricted, either by design or by nature (e.g. 'keyhole' channels).

Precoding is a common technique used in wireless transmissions where multiple data streams are spatially multiplexed by an access point or base station to a number of receiving stations by using multiple transmit antennas. The design of the precoder can be divided into two separate functional blocks: 1) the transmit filters, for example zero-forcing or minimum mean-square error, with unit norm columns and 2) the power loading. In a conventional system the transmit filters and power loading strategy are applied on individual OFDM subcarriers or groups of subcarriers. In this paper we study a codeword based power loading strategy combined with precoding. In this new technique the power allocation is done jointly across the subcarriers spanned by the encoded transmissions to optimise some codeword-related performance requirement. Simulations show that this strategy offers performance gain in terms of both packet error rate and throughput over conventional per-subcarrier power loading. The saving in computational complexity can also be significant, particularly for the MMSE precoder design.

In this letter, we introduce a novel method for demodulation of general amplitude- and phase-shift keying (APSK) constellations. By noting that the max-log log-likelihood ratio (LLR) is a piecewise linear function of the received signal, we devise a scheme to determine in which linear region that the received signal falls, obtain the linear parameters from a lookup table and compute the LLR. This method is general and works for any APSK (two-dimensional) constellation. By pruning the search tree, it also provides a mechanism for performance-complexity tradeoff and we show that its complexity is much lower than a previously published scheme (which only works for regular APSK constellations).

In this paper, we extend the concept of maximum-distance separable (MDS) array codes to a larger class of codes, where the array columns contain a variable number of data and parity symbols and the codewords cannot be arranged, in general, in a regular array structure with equal column length. These new codes, named irregular MDS array codes, find applications in problems of distributed data storage with multiple sources of information generating data at unequal rates. We solve the problem of finding optimal parity symbol allocations that achieve minimum redundancy for a given level of protection against block erasures. We provide a classification of irregular MDS array codes according to the parameters of their parity symbol allocation and we show how regular MDS array codes are a special case of this wider class. We derive necessary and sufficient conditions for such irregular array codes to be MDS and extend the concept of the lowest density generator matrix. Finally, we show how a simple constructive method allows to design irregular lowest density MDS array codes with alphabet size independent of the size of the array columns.

In this paper a switch-stay model is proposed for spectrum access by a secondary user in the presence of a primary user, which considers the impact of spectrum handoff delay. This model is then studied to balance the trade-off between the secondary user's energy efficiency and transmission reliability in light of the probability of collision, throughput and delay constraints. Bounds on the probability of stay on the current channel are derived from multiple constraints. An antenna selection sensing mechanism is then proposed to relax these bounds by enhancing the sensing performance. The advantages of the proposed antenna selection sensing are verified by numerical results.

Compressive sensing (CS) is a new signal acquisition framework, which allows for a signal recovery from far fewer samples than what is required by traditional sampling methods. In this paper we propose new strategies for adaptively adjusting the number of CS samples in wireless sensor networks (WSNs). Additionally, in the signal reconstruction procedure we apply homotopy algorithm to update the reconstructed signals. The reduction of CS samples and the homotopy update reduce the computational complexity and save processing time and energy for both the fusion centre and wireless sensors. The proposed techniques are investigated numerically in various WSN scenarios.

In this paper we derive a novel method for encoding the MIMO channel state information (CSI) for frequency selective channels and limited feedback. The technique is based on the decomposition of the directional information of a MIMO channel in a two-coordinate system formed by the tangent space of a Grassmann manifold and the tangent space of a unitary group. The first component provides a non-redundant description of the subspace spanned by a set of channel eigenvectors, while the second component provides a non-redundant addendum of information to identify the strongest channel vector directions within that subspace. One advantage of this feedback representation is that it allows to exploit the correlation of MIMO channels by applying standard source coding techniques to the two coordinate components. We also provide simulation results of an LTE downlink system, which show that the proposed CSI feedback description offers dramatic throughput improvement over the existing feedback scheme for MU-MIMO transmission.

We study the effect on diversity of using the sphere encoder-based vector perturbation (VP) technique of Peel and the lattice reduction-aided techniques (Babai approximations) of Windpassinger on a subset of the transmitted symbols. For a system withndiversity channels, we prove that the diversity order gap between the maximum diversity ofnachieved by the full-size VP and the minimum diversity of 1 provided by linear precoding can be bridged by applying VP to a group of transmitted symbols. In particular, we prove that the diversity order isn-l, wherelis the number of symbols that are not vector-perturbed. We show that this result holds true for lattice-reduction with both the rounding-off and nearest plane Babai approximations. The result shows that in terms of diversity, the group of vector-perturbed users counts as a single user. In fact, in a multiple-input and multiple-output Gaussian broadcast channel withllinearly precoded users, we can add up ton-lVP users and the diversity order for all users will decrease only by one, irrespective of how many VP users are added.

This paper presents a novel method for overlaying an analogue value on top of a digital signal. By adding a phase rotation and a signature sequence to a block of digitally modulated data symbols, this analogue value can be conveyed to a receiver without significant degradation to the underlying digital data. A near-ML estimator is derived which is non-data driven and hence independent of the digital data sequence. It is shown that this method of signalling a value is superior to quantising it and adding it to the digital data stream. We analyse the performance of the analogue and digital methods and also compare them using simulations.

In this paper we look at the problem of successively refining the description of a Gaussian i.i.d. source in a Grassmannian space. This problem is relevant, for example, in the encoding of channel state information for the limited feedback MIMO broadcast channel. We show how it is possible to achieve an isotropic distribution of the error vector by applying unitary transformations, which simplifies the problem of finding optimal codebooks for the error quantisation. Reconstruction error analysis and numerical tests show how the proposed technique can, in practice, perform very close to the distortion- rate bound and suggest that, in some cases, these successive error descriptions can be optimal refinements of one another.

Codebook-based limited feedback precoded MIMO systems involve vector quantization, which may require intensive computations, especially for large codebooks. In this paper we show how the complexity and codebook storage can be significantly reduced by using a special alphabet to generate the codebook. Simulation results show that the performance with the generated codebook is comparable with other codebooks which have higher complexity.

We propose a user selection algorithm and the supporting feedback scheme, for MIMO Gaussian broadcast channels with arbitrary number of users, transmit beamforming and limited channel state information (CSI) available at the transmitter. The method relies on estimation of the user signal-to-interference-plus-noise ratios (SINR's) at the transmitter, by means of an appropriate finite-rate channel report from the mobiles. The basic principle is that users are added successively only if they are beneficial to the system in terms of increased throughput. Numerical results show improved performance compared to other finite-rate feedback schemes.

Code-division multiple-access (CDMA) schemes based on block spreading implement the spreading of entire data blocks rather than single symbols, thus achieving a higher robustness against the frequency selectivity of the channel and allowing the use of efficient modulation/equalization schemes operating in the frequency domain (FD). In this paper, we present a new block CDMA (B-CDMA) system where a single cyclic prefix (CP) is used at the end of each spread block. This provides a higher spectral efficiency with respect to existing schemes. By observing that complete orthogonality among users is achievable only for half-loaded systems on dispersive channels, we introduce new criteria for the design of spreading and despreading sequences, which aim at minimizing the mean-square error at the output of the despreader. For the equalization of the received signal, we propose an iterative block decision feedback equalizer, which iterates between equalization and decoding. Equalization filters are designed to minimize the mean-square error and take into account the residual interference due to the nonorthogonality of the spreading sequences. The performance of B-CDMA is evaluated in an uplink wireless scenario and compared to existing CDMA schemes.

Dirty-paper coding (DPC) is a well-known construction capable of achieving the ultimate capacity limit of the MIMO broadcast channel (MIMO-BC). One recently proposed DPC scheme is based on Voronoi codes, minimum mean-square error generalised decision-feedback equaliser (MMSE-GDFE), dither and lattice decoding. Building on this approach, in this paper we introduce a linear pre-coder for the MIMO-BC, which, although sub-optimal compared to the original DPC construction, may be regarded as a valuable option for practical applications because of its simplicity and its close-to-capacity achievable rates. We provide a theoretical analysis as well as numerical results which show the achievable rates of our linear pre-processing scheme compared to the DPC construction and to another recently proposed linear technique for throughput maximisation, based on gradient nulling.

In this paper we propose a new multiple user architecture for broadband wireless communications based on code division multiple access, denoted block CDMA (B-CDMA). The B-CDMA transmitter spreads the data on a block basis and a single cyclic prefix is used at the end of each spread block. We investigate the orthogonality of spreading sequences for B-CDMA, for an uplink broadband transmission and we design sequences that minimize the power of multiuser access interference. Equalization of the received signal is performed through an iterative block decision feedback equalizer (IBDFE). The B-CDMA system is compared to existing block spreading schemes in terms of achievable throughput for a wireless broadband uplink transmission

This letter focuses on maximization of achievable throughput for uplink orthogonal frequency-division multiplexing code-division multiple-access (OFDM-CDMA) systems. In particular, a new pre-equalization technique is presented where all users employ spreading both in frequency (on sub-bands of the available spectrum) and time; moreover, bit and power loading is used at the mobile terminals to accomplish throughput maximization. To avoid interference among users, the proposed loading algorithm inherently inverts the channel amplitude on the sub-carriers of each sub-band, while the phase is adjusted separately.

In this paper we address the problem of how best selecting the smallest set of transmission modes (i.e. modulation and channel code formats) to be employed in adaptive modulation and coding (AMC) schemes for OFDM. In fact, by decreasing the number of allowable modes, the complexity of the ACM scheme is greatly reduced. The method makes use of an optimum throughput maximization algorithm for the bit and power allocation. We start with a large set of possible transmission modes. Then, for a given channel, we iterate the loading algorithm with mode sets having successive removal of the mode which yields the smallest contribution to the overall throughput. By using this method in a Rayleigh fading channel, we are able to show that in the test case of the HIPERLAN/2 modes, up to five modes out of seven can be removed with a very small impact on the average achievable throughput.

A comparison in the performance between hard- and soft-decision Viterbi decoding, with application to the HIPERLAN/2 standard, is presented. The results show that when channel state information (CSI) is included in the generation of the soft-decision information, the soft-decision method greatly outperforms the hard method. Moreover, a simplified algorithm for the soft-output demapper for the 16-QAM and 64-QAM constellations is developed, which allows the complexity of the demapper to be maintained at almost the same level for all the possible modes of HIPERLAN/2.