Rouzbeh Razavi

Self-optimization of capacity and coverage in LTE networks using a fuzzy reinforcement learning approach

This paper introduces a solution to enable self-optimization of coverage and capacity in LTE networks through base stations downtilt angle adjustment. The proposed method is based on fuzzy reinforcement learning techniques and operates in a fully distributed and autonomous fashion without any need for a priori information or human interventions. The solution is shown to be capable of handling extremely noisy feedback information from mobile users as well as being responsive to the changes in the environment including self-healing properties. The simulation results confirm the convergence of the solution to the global optimal settings and that the proposed scheme provides up to 20% performance improvement when compared with an existing fuzzy logic based reinforcement learning approach.

A Fuzzy reinforcement learning approach for self-optimization of coverage in LTE networks

Optimization of antenna downtilt is an important aspect of coverage optimization in cellular networks. In this paper, an algorithm based on the combination of fuzzy logic and reinforcement learning is proposed and applied to the downtilt optimization problem to achieve the self-configuration, self-optimization, and self-healing functionalities required for future communication networks. To evaluate the efficiency of the proposed scheme, we use a detailed Long Term Evolution (LTE) simulation environment and employ an algorithm for configuring and optimizing the downtilt angle of the LTE base station antennas. This scheme is fully distributed and does not require any synchronization between network elements. Compared to an existing solution, evolutionary learning of fuzzy rules (ELF), the solution we propose provides up to 20 percent improvement in performance. In addition to self-x capabilities, the experiments further confirm the reliability and robustness of the algorithm in extremely noisy environments.

Urban small cell deployments: Impact on the network energy consumption

This paper investigates the effect of small cell deployments in urban and dense urban areas from the energy consumption perspective. Considering a parametric power model for the legacy macrocell networks and the new emerging small cells, this study quantifies the power reduction gain by deploying heterogeneous networks consisting of a mixture of both technologies. As a part of the study, a framework is developed which determines the optimal network architecture in terms of the combination of small cells and macrocells. While the optimality is merely decided based on the overall power consumption performance, both capacity and coverage requirements are simultaneously addressed. The results presented in this paper are based on the field measured traffic demand data from the urban area of the Wellington, NZ. Additionally, the paper investigates the impact of future traffic growth and provides a 5-year outlook for energy consumption of the network. The presented numerical results confirm substantial power reduction gain from deploying small cells. This becomes even more critical in the future when the traffic demand is increasing and the offloading effects of small cells are most beneficial. Furthermore, this study suggests that improving the idle mode power consumption of small cells is one of the key areas which can enable significant total power reduction. This study can serve as a guideline for operators when estimating the power efficiency of their network.

H.264 video streaming with data-partitioning and Growth codes

This paper demonstrates that Growth codes, based on Raptor channel coding, allow incremental protection of H.264 video codec data-partitioned Network Adaption Layer units. When combined with increased protection of video reference frames, in an ADSL erasure channel up to 10 dB in video quality (PSNR) can be gained through this scheme compared to equal error protection with rateless codes. Equivalent gains occur in a wireless channel from combining data-partitioning with error protection. The bitrate overhead from data-partitioning is also shown to be less than from other H.264 error resilient tools.

Adaptive packet-level interleaved FEC for wireless priority-encoded video streaming

Packet-level Forward Error Control (FEC) for video streaming over a wireless network has received comparatively limited investigation, because of the delay introduced by the need to assemble a group of packets. However, packet-level interleaving when combined with FEC presents a remedy to time-correlated error bursts, though it can further increase delay if this issue is not addressed. This paper proposes adapting the overall degree of interleaved packet-level FEC according to the display deadlines of packets, transmit buffer occupation, and estimated video input to the wireless channel, all of which address the issue of delay. To guard against estimation error, the scheme applies a conservative adaptation policy, which accounts for picture type importance to ensure that display deadlines are met, thus avoiding this defect of interleaving. The paper additionally introduces a greedy algorithm that effectively groups packet-level FEC protection according to packet priority. Priority encoding adds extra protection during deep fades. As feedback is not required, the interleaving scheme is suitable for all forms of video broadcast. A Bluetooth piconet demonstrates the packet-level FEC interleaving scheme, which provides higher quality delivered video compared to the industry-standard Pro-MPEG Cop#3r2 interleaving scheme.

Utility fair optimization of antenna tilt angles in LTE networks

We formulate adaptation of antenna tilt angle as a utility fair optimization task. This optimization problem is nonconvex, but in this paper we show that, under reasonable conditions, it can be reformulated as a convex optimization. Using this insight, we develop a lightweight method for finding the optimal antenna tilt angles, making use of measurements that are already available at base stations, and suited to distributed implementation.

Self-Optimisation of Antenna Beam Tilting in LTE Networks

This paper introduces a novel technique for self-optimisation of antenna down-tilting. The proposed solution consists of two phases. During the first phase, which needs to be performed quickly, a method based on the Golden Section Search (GSS) algorithm is used to find a near-optimal tilt angle. Then, in the next step, the tilt angle value is fine-tuned through frequent explorations of the near-optimal region. Additionally, the solution benefits from self-healing properties by responding to the failure of a neighbouring cell. In this case, the algorithm is adapted to find a new optimal point upon failure detection. The result shows fast convergence and close-to-optimum performance (93.8% on average) of the proposed algorithm. The overall solution is fully distributed and is 3GPP complaint.

Multiconstraints fuzzy-logic-based scheduling algorithm for passive optical networks

Presented is a fuzzy-logic-based scheduling algorithm for passive optical networks (PONs) that considers four different metrics to allocate an upstream bandwidth to optical network units (ONUs). The metrics considered are the delay of the head-of-the-line packets at the ONUs, the importance level of the packets, the relative ONUs buffer fullness, and the level of the power fluctuation from one ONU to another. One of the advantages of a fuzzy controller is the fact that, regardless of the design complexity, the controller can be implemented as a simple look-up table, which makes it ideal for high-speed operation. Further facilitation of implementation was achieved by realization of the fuzzy algorithm through a two-stage hierarchal architecture. Moreover, linear predictive filters have been used to predict the traffic arrival rate and the packet delay at the ONUs. Compared with the round-robin scheduling algorithm, the results show significant performance improvement in terms of the overall packets delay as well as jitter when the proposed algorithm in employed. Furthermore, using this algorithm would reduce the average level of power fluctuations in a PON system and will also provide high-level service differentiation between packets of different importance.

Fuzzy logic control of adaptive ARQ for video distribution over a bluetooth wireless link

Bluetooths default automatic repeat request (ARQ) scheme is not suited to video distribution resulting in missed display and decoded deadlines. Adaptive ARQ with active discard of expired packets from the send buffer is an alternative approach. However, even with the addition of cross-layer adaptation to picture-type packet importance, ARQ is not ideal in conditions of a deteriorating RF channel. The paper presents fuzzy logic control of ARQ, based on send buffer fullness and the head-of-line packets deadline. The advantage of the fuzzy logic approach, which also scales its output according to picture type importance, is that the impact of delay can be directly introduced to the model, causing retransmissions to be reduced compared to all other schemes. The scheme considers both the delay constraints of the video stream and at the same time avoids send buffer overflow. Tests explore a variety of Bluetooth send buffer sizes and channel conditions. For adverse channel conditions and buffer size, the tests show an improvement of at least 4 dB in video quality compared to nonfuzzy schemes. The scheme can be applied to any codec with I-, P-, and (possibly) B-slices by inspection of packet headers without the need for encoder intervention.

Self-configuration of scrambling codes for WCDMA small cell networks

This paper introduces the problem of Primary Scrambling Code (PSC) selection in small cell networks and proposes a novel optimisation technique. Small cells introduce challenges not present in conventional macrocell scrambling code allocation, including the need for dynamic allocation, scalable distributed allocation algorithms, and support for unplanned and organic deployments. To the best of our knowledge this is the first study addressing the issue of distributed scrambling code selection for small cell networks. We propose a decentralized learning algorithm which does not require any collaboration between the neighbouring base-stations and which finds a feasible allocation whenever one exists. The performance of the algorithm is compared against two variations of a greedy algorithm which is the current 3GPP recommendation and is shown to offer significant performance benefits.