My research interests are in the broad areas of wireless communications. In particular, I'm interested in algorithm development and signal processing for wireless communication systems, including providing connectivity to massive IoT, network densification and massive MIMO for 5G networks, and dynamic spectrum access. A list of research directions that span my interests is shown below.
My research interests span the following areas
- Massive IoT communications: cellular-based and unlicensed-based solutions
- User association and load balancing in ultra-dense cellular networks
- Resource allocation and interference management for 5G networks
- Spectrum sensing and processing for dynamic spectrum access
- Wireless sensor networks
currently working on developing solutions for massive IoT communications. Specifically, we aim to propose both cellular-based and unlicensed-based solutions that connect a large number of machines and sensors to the Internet. In the cellular-based one, we propose to complement cellular networks with drones to aggregate IoT data to the cellular infrastructure. In the latter, we aim to develop a distributed wideband sensing procedure that identifies white spaces at a finer spatial scale, which suits the small footprint of IoT devices.
In addition, I have worked on two problems pertaining 5G networks: load balancing and interference management. Specifically, we have developed a user-centric association strategy that exploits the massive MIMO capabilities of macro BSs to maximize the long-term throughput of the user in cellular networks with spectrum sharing. To tackle inter-tier interference, we have further developed two maximization frameworks with joint cell range expansion and spectrum allocation: rate-based formulation for human-type traffic and coverage-based formulation for machine-type traffic.
During my master's degree, my research has centered around dynamic spectrum access. Specifically, we investigated the challenges and limitations of state-of-the-art multiband spectrum sensing techniques, proposed a pragmatic cooperative sensing framework for multiband sensing, and investigated the benefits of network-wide based reconfiguration in both centralized and distributed cognitive radio networks. In addition, we have proposed new sensing algorithms, including an enhanced pilot-tone aided detector that overcomes imperfect phase synchronization and noise power uncertainty and a confidence-based generalized combining scheme for cooperative sensing.
During my bachelor's degree, I have worked on the design of an underwater wireless sensor network (WSN). In particular, we developed a realistic channel model to understand the path loss in underwater environments, and designed a prototype that enables WSNs to use electromagnetic waves as means of communications.
Finally, I have worked on a personal project, where I have written a book on the fundamentals of signal detection and estimation.
With the proliferation of wireless devices and the advent of bandwidth hungry applications such as video streaming, cloud-based technologies, etc., incremental improvements to existing networks and architectures such as 4G are insufficient to meet the projected data demands in the next five years. This has called for paradigm shifts in next generation networks, i.e., 5G, and particularly, the ongoing evolution towards very dense and unplanned deployment of low-power small cell base stations (BSs) of various types, commonly known as HetNets.
Small cells can significantly enhance network’s throughput, but there are several key challenges associated with the deployment of HetNets such as load-balancing and the high interference. For instance, low-power BSs can be overshadowed by the high transmit power of the macro BS, rendering existing user association techniques such as the max-power user association impractical due to the load-imbalance it causes in the network. Similarly, co-channel deployment of these cells prohibitively increases the interference, degrading the quality-of-service especially for users located at cell edges.
We propose load-aware user association
policies that significantly improve the user’s throughput by balancing the load
for different network settings. In addition, we study interference not only via
the allocation of time-frequency resources but also via the allocation of
spatial resources based on the premise that massive MIMO will be an integral
technology in 5G.
Illustration of a cellular network with multiple tiers, i.e., base stations across different tiers have different coverage areas and capabilities.
There are available spectral opportunities (known as white spaces) in time, frequency, and space. Cognitive radio promises to utilize these opportunities.
The primary task of a spectrum sensor is to determine if a spectral opportunity is available or not, i.e., the SU must determine if a channel is occupied by a PU or not. Significant milestones have been achieved in single-band spectrum sensing, yet the throughput of the single-band spectrum access scheme is limited and frequent data transmission interruptions are inevitable. The latter is due to the fact that the SU must periodically sense the channel, after accessing it, to account for sudden reappearances of the PU. Consequently, these limitations have motivated the need for multiband spectrum access, where the SU senses multiple channels at the same time. We have shown in our work that this paradigm significantly improves the network's throughput. More importantly, it can provide better link maintenance since the SU can seamlessly handover from one channel to another.
In addition to sensing a wide swath of the spectrum, sensing each channel generally requires multiple SUs to collaborate and share their spectrum sensing measurements/results. Such cooperative processing of measurements is critical to overcome the hidden terminal problem and provide diversity against the different channel impairments. Existing work on cooperative spectrum sensing for multiband CRNs has focused on the case where each SU senses the entire spectrum. This makes information exchange very expensive even if one-bit decisions, i.e., hard decisions, are exchanged due to the large overhead incurred when multiband sensing is used. Equally important, multiband sensing demands advanced receiver front-ends. These challenges have motivated us to make a compromise. Particularly, it is more practical if each SU senses a subset of the spectrum (instead of the entire spectrum) such that all SUs sense the entire spectrum at the end. This interesting compromise leads to a fundamental tradeoff when cooperative communications is used in the case of multiband CRNs. Particularly, there is an inherent tradeoff between the diversity and the sampling complexity. For instance, to reduce the sampling complexity, the SU must senses fewer channels. This means that for a given channel, fewer SUs will be sensing it, i.e., the diversity is decreased.
In this figure, it is assumed that there are six SUs. Conventionally, each SU must sense the entire spectrum, and thus each channel is sensed by six SUs (i.e., full diversity). However, this comes at the expense of large overhead and very high sampling rate. To mitigate these challenges, each SU can sense a subset of the spectrum. In (a), non-uniform diversity is achieved. For example, some channels are sensed by one SU whereas other channels are sensed by multiple SUs (based on a predetermined criterion). In (b), uniform diversity of two is achieved since each channel is sensed by two SU.
This figure illustrates the fundamental tradeoff between diversity and the sampling rate. It is observed that the higher the diversity, the higher the sampling rate requirement. Also, the more channels, M, the SU senses, the higher the sampling rate. Observe that if full diversity is desired (i.e., diversity of K), then the sampling cost is invariant of K. For instance, when M=12 and K=2, then the full diversity is two, and each SU will have to sense 12 channels. Now let K=10. Then, the full diversity is ten, and to achieve it, each SU must sense 12 channels. Thus, the sampling cost remains the same regardless of the number of cooperating SUs.
Staff: Mohamed Ibnkahla (PI) and Ghaith Hattab
- G. Hattab and M. Ibnkahla, "Multiband Spectrum Access: Great Promises for Future Cognitive Radio Networks," Proceedings of the IEEE, Mar. 2014.
- G. Hattab and M. Ibnkahla, "Multiband Spectrum Sensing: Challenges and Limitations", Proc. WiSense Workshoptawa, August 2014.
Feature-based spectrum sensing techniques compromise a balance between energy and coherent detectors. The former requires minimal prior information of the observed signal, and the latter has robust detection performance when the observed signal is very weak. In this project, we have focused on pilot-based feature detectors. We have proposed an improved pilot tone-aided detector (PTD) that intelligently utilizes both the pilot tone and the energy of the received signal, and we have shown that this detector is optimal in the Neyman-Pearson sense.
This figure shows that the proposed detection technique (the improved PTD (IPTD)) outperforms the existing techniques (the coherent-based (CPTD) and the energy-based (EPTD)) for different SNR values (taken from ).
Staff: Mohamed Ibnkahla (PI) and Ghaith Hattab
- G. Hattab and M. Ibnkahla, ""Enhanced Pilot-Based Spectrum Sensing Algorithm," Proc. IEEE Biennial Symps. on Commun. (QBSC'14), June 2014.
Primary systems require protection from secondary systems even in worst-case scenarios. Hence, sensors must reliably sense the spectrum in poor channel conditions, e.g., a sensor in deep fade. For this reason, it is desired to perform sensing in a cooperative manner, i.e., multiple sensors share their measurements/local decisions with others to arrive at one global decision regarding the occupancy of the channel.
There are a plethora of algorithms and techniques to combine/fuse the sensing measurements of multiple sensors. In our work, we focus on centralized fusion and processing of quantized energy measurments. The quantization procedure maps the energy measurement into a confidence-level using a fuzzy-logic membership function. Such approach strikes a balance between hard-combining, i.e., processing of one-bit local decisions, and soft-combing, i.e., processing of actual local measurements.
In addition to this work, we have worked on a cooperative sensing framework where different sensors have different sensing capabilities, e.g., some sensors have feature detectors and others have energy detectors.
This figures illustrates the performance of the proposed cooperative sensing framework (denoted as JQCQ) in comparison with hard-combining and soft combining.
Staff: Mohamed Ibnkahla (PI), Ghaith Hattab (Co-PI), Waleed Ejaz, Nesrine Cherif, and Takwa Attia
- W. Ejaz, G. Hattab, T. Attia, M. Ibnkahla, F. Abdelkefi, and M. Siala, "Joint Quantization and Confidence-based Generalized Combining Scheme for Cooperative Spectrum Sensing," to appear in IEEE Systems Journal, Sept. 2016.
- W. Ejaz, G. Hattab, N.
Cherif, M. Ibnkahla, F. Abdelkefi, and M. Siala, "Cooperative Spectrum
Sensing with Heterogeneous Devices: Hard Combining versus Soft Combining," to appear in IEEE Systems Journal, June 2016.
Most of the existing technologies for underwater communications rely on acoustic waves as a transmission medium because they can propagate for very long distances in deep water. However, for near-shore applications, such as monitoring the swimmers and coastal surveillance, acoustic-based networks fall short because they are severely affected by the ambient noise. Another limitation is that these waves have very low speed, compared to electromagnetic waves, and thus they support very low data rates. As a result, most of the near-shore applications are implemented using cables, which are not only expansive to implment but also require regular maintenance. For instance, in one of our meetings with Dubai Coastal Zone Monitoring engineers, they have shown us a cable-connected radar network near the shore. They have explained that storms usually damage the cables, and if they are not damaged, these cables must be calibrated again, which is a frustrating procedure.
Staff: Mohamed El-Tarhuni (PI), Nasser Qaddoumi (Co-PI), Ghaith Hattab, Moutaz Al-Ali, and Tarek Joudeh
- G. Hattab, M. El-Tarhuni, M. Al-Ali, T. Joudeh, and N. Qaddoumi, "An Underwater Wireless Sensor Network with Realistic Radio Frequency Path Loss Model", International Journal of Distributed Sensor Networks, Mar. 2013.
- G. Hattab, M. Faisal, and T. Joudeh, "Design of Underwater Wireless Sensor Network (UWSN)", Technical Report, May 2012.---