One of the main challenges keeping flash memories from achieving widespread distribution is their limited endurance. The programming and erasing from re-writes damages the cells, progressively increasing the number of errors until information can no longer be stored reliably. Most manufacturers employ powerful ECC techniques, but there is a limit to the number of errors that these can correct. When the number of errors goes beyond the capability of the ECC, it is necessary to invoke RAID, which introduces significant latency and jeopardizes the speed of the memory if used too often. This paper introduces a method for estimating the BER that a flash page will exhibit after retention and uses this estimate for wear leveling. Instead of leveling out the number of PE cycles in all the blocks, the proposed scheme attempts to wear all the blocks evenly so that they all suffer the same BER. Additionally, the estimate will be used to detect bad blocks, those likely to exhibit a number of errors beyond the ECC correction capability, and retire them from further use.

One of the main attractions for the widespread of flash memories in data storage is their fast read speed. Read speed can potentially be jeopardized when there are a large number of read errors as a result of flash cell wear-and-tear, leakage, and many other factors. Powerful (error-correction coding) ECC techniques are generally employed, but the number of errors that these schemes can correct is limited; therefore, often other sophisticated error correction mechanisms such as RAID need to be invoked, as well, which introduce significant latency. Flash pages that are not ECC correctable are identified as bad pages. By proactively reading pages, decoding the data and comparing the bit-error with a pre-fixed bit-error threshold value one can potentially predict bad pages and remove their content before they become problematic. This paper proposes finding such bit-error thresholds by formulating the proactive read process as a Markov Decision Process (MDP). Based on Dynamic Programming (DP) an algorithm can then be obtained that selects a maximum bit-error value (threshold) above which pages should be removed. Such threshold values can simply be stored as a look-up table in the controller ahead of time.

Two important performance metrics for a storage system are the latency associated with retrieving data from its storage medium and the effective lifetime of its storage medium. Both metrics are directly affected by the number of raw read errors (i.e. errors prior to exploiting error-correction mechanisms). This paper focuses on NAND flash memories, where a read is performed by comparing stored voltages with a threshold voltage. The unwanted variation of stored voltages causes read errors. This paper identifies number of flash program-erase (PE) cycles, time elapsed between writing and reading, and the page number (physical location) as the main sources of voltage variations. It then proposes a method for learning how read thresholds should vary with these parameters such that the storage controller can dynamically vary thresholds and minimize read errors. Lab experiments show that at the flash end-of-life, the proposed method lowers the raw bit-error-rate up to a factor of 6, as compared to manufacturer's default read settings.

A primary source of increased read time on NAND flash comes from the fact that, in the presence of noise, the flash medium must be read several times using different read threshold voltages for the decoder to succeed. This paper proposes an algorithm that uses a limited number of rereads to characterize the noise distribution and recover the stored information. Both hard and soft decoding are considered. For hard decoding, this paper attempts to find a read threshold minimizing bit error rate (BER) and derives an expression for the resulting codeword error rate. For soft decoding, it shows that minimizing BER and minimizing codeword error rate are competing objectives in the presence of a limited number of allowed rereads, and proposes a tradeoff between the two. The proposed method does not require any prior knowledge about the noise distribution but can take advantage of such information when it is available. Each read threshold is chosen based on the results of previous reads, following an optimal policy derived through a dynamic programming backward recursion. The method and results are studied from the perspective of an SLC Flash memory with Gaussian noise, but this paper explains how the method could be extended to other scenarios.

On NAND flash, a primary source of increased read time comes from the fact that in the presence of noise, the flash medium must be read several times using different read threshold voltages to find the optimal read location, which minimizes bit-error-rate. This paper proposes an algorithm to estimate the optimal read threshold in a fast manner using a limited number of re-reads. Then it derives an expression for the resulting BER in terms of the minimum possible BER. It is also shown that minimizing BER and minimizing codeword-error-rate are competing objectives in the presence of a limited number of allowed re-reads, and a tradeoff between the two is proposed.

The aggressive scaling of the NAND flash technology has led to write noise becoming the dominant source of disturbance in the currently shipping sub-30 nm MLC NAND memories. Write noise can be mitigated by reducing the magnitude of the voltage levels programmed into the cells, which additionally translates to longer flash memory lifetime. However, if all the target levels are small and close together, the probability of error could become excessively high. It is therefore necessary to optimize the target level placement in order to achieve a trade-off between flash lifetime and error probability. This paper proposes a method to maximize flash lifetime subject to reliability constraints, and vice versa. Simulation results show that the proposed method doubles flash lifetime in comparison to a naive scheme, for a 2% reliability constraint. It also comes very close to the optimal solution obtained by brute force search, while maintaining negligible computational complexity in comparison.

With the advent of new applications, one of the primary applications supported in downlink cellular systems is streaming video. QoS for streaming video entails data transmission to the mobile stations with limited latency and limited jitter. This requirement on latency and jitter can be mapped to the number of yet-to-be-played packets in a mobile station's buffer, to always be above a certain minimum. Additionally, there should be no overflow of finite-size user buffers. This paper solves the problem of optimal resource allocation in a downlink OFDMA system with buffer constraints for all downlink users. The optimal strategy is shown to be a generalized version of water-filling and channel-inversion. Specifically, with infinite buffer size and no minimum constraint on buffer, it reduces to water-filling; and to channel-inversion in the absence of a buffer. It is shown that scheduling gains from the derived optimal strategy, in terms of number of users that the system can support, can be an order-of-magnitude higher than that of heuristic scheduling schemes.

This paper studies linear transmit filter design for weighted sum-rate (WSR) maximization in the multiple input multiple output broadcast channel (MIMO-BC). The problem of finding the optimal transmit filter is non-convex and intractable to solve using low complexity methods. Motivated by recent results highlighting the relationship between mutual information and Minimum Mean Square Error (MMSE), this paper establishes a relationship between weighted sum-rate and weighted MMSE in the MIMO-BC. The relationship is used to propose a low complexity algorithm for finding a local weighted sum-rate optimum based on alternating optimization. Numerical results studying sum-rate show that the proposed algorithm achieves high performance with few iterations.

In the OFDMA downlink, obtaining channel state information (CSI) from users is necessary for the base station (BS) to optimize network performance by intelligently allocating resources and scheduling mobile stations (MS). However, the overhead of obtaining CSI is a large burden and therefore, schemes to reduce CSI are necessary for a realizable system. By considering a MS's coherence time Deltatcohand coherence bandwidth Bcoh, and exploiting this redundancy in the MS's CSI, the amount of feedback can be tailored to the specific user's profile and greatly decreased. Typically, reducing feedback results in more uncertainty in CSIT and performance degradation. However, in the proposed scheme, the CSIT's deviation from the true channel state is bounded, and thus can provide robust scheduling. Specifically, two BS schemes are suggested that either fix the average BER or the average outage probability regardless of user mobility or delay spread.

This paper studies the downlink Orthogonal Frequency Division Multiplexing (OFDM) setup with a single Base Station (BS) serving many users. The BS is assumed to have limited Channel State Information (CSI) obtained by explicit feedback from the downlink users. Given that the feedback rate and the coherence time of the channel are fixed, the question asked in this paper is: how to allocate the feedback resources optimally? Specifically, what is the optimal number of tones grouped as a subchannel, the number of users that feedback for any subchannel and the number of bits used for quantization of CSI? Analytical expressions are derived for the i.i.d. Rayleigh fading case and it is shown that there is a definite hierarchy in the importance of the three design variables. Feedback resources are first allocated to create the maximum number of subchannels possible, then to allow for more users to feedback for any subchannel and lastly to increase the precision of the channel value. Monte-Carlo simulations are performed to verify the accuracy of the derived analytical expressions.

Optimal tone allocation in downlink OFDMA networks is a non-convex NP-hard problem that requires extensive feedback for channel information. In this paper, two constant complexity limited-feedback algorithms are proposed to achieve near-optimal performance. First, using opportunistic feedback, the proposed schemes are shown to reduce feedback overhead by requiring only users likely to be allocated resources to feedback. There are differences between the two proposed schemes for implementation of the feedback protocol. One scheme requires less feedback but is contention-based, while the other scheme is sequential and thus avoids possible collisions leading to slightly higher performance, but needs more feedback. Second, complexity is reduced for resource allocation by solving the optimization problem in a distributed manner, rather than centrally at the base station. As shown both analytically and through numerical results, these distributed algorithms reduce the required feedback overhead significantly, and achieve constant computational complexity with little performance loss compared to the optimal solution.

This paper studies linear transmit filter design for Weighted Sum-Rate (WSR) maximization in the Multiple Input Multiple Output Broadcast Channel (MIMO-BC). The problem of finding the optimal transmit filter is non-convex and intractable to solve using low complexity methods. Motivated by recent results highlighting the relationship between mutual information and Minimum Mean Square Error (MMSE), this paper establishes a relationship between weighted sum-rate and weighted MMSE in the MIMO-BC. The relationship is used to propose two low complexity algorithms for finding a local weighted sum-rate optimum based on alternating optimization. Numerical results studying sum-rate show that the proposed algorithms achieve high performance with few iterations.

This paper presents an opportunistic feedback (OF) protocol that reduces the amount of channel-state-information (CSI) feedback for the OFDMA downlink system. The OF protocol associates a channel value instead of a user to each feedback slot. The users access the slots opportunistically based on their channel state measurement for each tone. This allows the base station (BS) to deduce the channel state from the slot index without explicit channel-state feedback, and also solve the resource allocation problem on the fly. The OF protocol is designed to support heterogeneous traffic, where users with deadline-sensitive (DS) applications are prioritized over best- effort (BE) users. The proposed OF protocol (a) requires finite number of feedback slots upper bounded by a small number, (b) is fully distributed, and (c) provides QoS to the DS applications.

We consider optimal allocation of resources to users in a downlink OFDMA system to support heterogeneous applications consisting of both deadline-sensitive (DS) and best-effort (BE) data in a cross-layer manner. Given the user queue-states and long-term channel statistics, the proposed persistent scheduling algorithm allocates the minimum resources to 'just' meet the deadlines for DS packets (QoS guarantee). The remaining resources are optimally split between the users for their BE data to maximize sum-rate, minimize buffer holding cost or any general utility function. We formulate the resource allocation problem as a single geometric program (GP) that can be solved using standard convex optimization software tools. Simulation results are presented to compare the performance of different objectives for the BE applications in presence of DS traffic.

This paper considers a wireless uplink, where some users have deadline-sensitive (DS) traffic and others have best- effort (BE) traffic. The objective is to meet the deadlines for DS packets a fixed percent of time (QoS guarantee), at the same time minimizing the sum-power used for transmission in a distributed manner without feedback fromallusers to the base station (BS). Given each user's channel state and rate requirement, the optimal user ordering that minimizes sum-power is shown to be independent of the rate requirements and depends solely on the channel conditions. Based on this, using a limited amount of feedback, the proposed opportunistic feedback (OF) protocol guarantees QoS to DS data and minimizes sum-power. Numerical results show that (i) the OF protocol provides deadline-meeting guarantees and (ii) minimizes sum-power with a greatly reduced feedback overhead as compared to that of a central scheduler.

We consider a MIMO broadcast channel (BC) with large M and K, where M is the number of transmit antennas and K is the number of single-antenna users; and propose a scalable feedback protocol that achieves the sum-capacity with limited feedback of channel state information (CSI). In our earlier work, we showed that if feedback time slots correspond to channel gains and not to users, the sum-capacity M log log K can be achieved with feedback resources growing only as M log K, unlike linearly as M K for conventional schemes. In this work, we show that feedback requirement can further be reduced by half or more, while still achieving the sum-capacity. A scalable feedback (SF) protocol using random beamforming (RBF) is proposed, which a) requires finite number of feedback slots upper bounded by a constant, b) is fully distributed, c) needs finite transmission energy during feedback and d) achieves the sum-capacity. Numerical results show that feedback load is substantially reduced as compared to conventional schemes, as well as our own previously proposed SF scheme.

This paper presents an opportunistic feedback (OF) protocol that achieves the asymptotic sum-capacity of the fading broadcast channel (BC) with a limited amount of feedback. The OF protocol associates a channel value instead of a user to each feedback slot. The users access the slots opportunistically based on their channel state measurement. This allows the base-station to deduce the channel state from the slot index without explicit channel-state feedback. This protocol is then extended to the MIMO BC by using zero-forcing beamforming and random beamforming. The OF protocol is analytically shown to (a) require finite number of feedback slots upper bounded by a small number, (b) be fully distributed, (c) need minimal transmission energy during feedback and (d) asymptotically achieve the sum-capacity of the MIMO BC. Comparison with other opportunistic feedback schemes shows that the proposed scheme has substantially larger sum-rate with a lower feedback overhead.

This paper studies optimal allocation of resources to users in a fading broadcast channel (BC) to support heterogeneous traffic. Resource allocation is done in a cross-layer manner for both deadline-sensitive (DS) and best-effort (BE) data to provide users guaranteed Quality of Service (QoS). Given the user queue-states, the proposed algorithm allocates the minimum resource to 'just' meet the deadlines for DS packets a fixed % of time. The remaining resources are then optimally split between the users for their BE data to maximize sum-rate, minimize buffer holding cost or any general utility function. The resource allocation problem is formulated as a linear program (LP) followed by a geometric program (GP) which can be solved very efficiently. Simulation results are presented for Poisson packet arrivals and exponentially distributed packet lengths on a Rayleigh fading channel for both the DS and BE traffic.

Geolocation using received signal strength (RSS) has large errors due to multipath fading, since fading results in high variations in RSS. We show how and when spatial diversity combined with channel knowledge at the receiver can be used to combat fading effects to increase accuracy in location estimation. We then propose a simple scheme for distance estimation, characterize the channels for which improvement in distance estimates can be thus obtained and prove that the mean square error of the distance estimate converges to zero with increasing diversity order. It is observed that the improvement can always be obtained for the Rayleigh channel, and for the Nakagami-m channel if the parameter m remains the same regardless of the distance.