Research

My research interests lie in signal processing and wireless communications. Below are some of the research areas I have worked or am currently working on.

User Association in D2D and Massive MIMO Heterogeneous Networks (current)

Users will have many options for communications in the future, so which option will be best? Small cells should take precedence when it comes to supporting users, but D2D and massive MIMO have advantages also.


An important consideration in determining user association is fairness and load balancing. It is undesirable to have one cell or base station handle too much load compared to others, as this limits future users' options and current users' resource allocations.


In particular, balance and fairness are not interchangeable terms, as one can envision a scenario where all users may receive the same rate from the network (and hence experience total fairness), but under-used base stations means that the network as a whole is imbalanced. We can construct a metric to capture this notion of network balance, and use it to help refine user association.


Another interesting question regarding user association is whether subsequent user associations can be predicted or estimated instead of recalculated when the network undergoes a small change, e.g., one user enters or leaves. If so, this can save computation time if the number of users is large.


Mode Selection and Resource Optimization for D2D Underlaying Heterogenous Networks

Device-to-device (D2D) communications is a promising communications technology that can improve efficiency, system capacity and reliability in future communications systems. A key question regarding D2D is which mode potential D2D devices should communicate in given current network states.


The three common D2D modes are:

  • Direct D2D - a potential D2D pair communicates direct to each other using dedicated frequencies.
  • Reuse - D2D pair communicates directly to each other, but shares the same frequencies as other cellular and/or small cell users.
  • Cellular - D2D is not permitted (e.g., due to interference limitations), and potential pairs are treated as normal cellular users.
    • There have been mode selection algorithms and critiera investigated in literature, but D2D underlaying heterogeneous networks is still an emerging research area. In particular, resource allocation considerations (e.g, power and/or frequency allocation) must be jointly made with small cells, macro base stations, and device transmitters.


    Precoding for Interference Suppression in Heterogenous Networks

    The existence of heterogeneous networks and small cells creates interference problems for all users located within overlapping cell regions. One method of combating interference on the downlink is through precoding. In particular, a generalized inverse precoder at a macro base station can be used to determine the amount of interference suppression targeted towards a specific user. Such a strategy maintains interference free transmission to multiple cellular users, and can improve overall user fairness.


    Alamouti Coding for MIMO Relay Networks (Honours thesis)

    Alamouti coding is used for various MIMO systems to improve diversity and overall reliability. In relay systems, Alamouti coding can be applied to both the transmitter and relay to ensure maximum diversity across all links.


    However, spatial correlation (e.g. from the small spacing between multiple antennas) reduces the independence of the antennas, and thus can limit the capacity of a wireless system. The pairwise error probability of an Alamouti coded relay system depends on the correlation matricies of the transmitter, relay and receiver nodes.


    Spectrum Sensing for Cognitive Radio

    Cognitive radio has become a promising solution to combat frequency congestion by ubiquitious wireless devices. Spectrum sensing is the first process by which a secondary user determines whether it should use frequencies dedicated to primary users.


    Various spectrum sensing methods have been proposed in literature, including those based on energy, eigenvalues and equal gain assumptions. Other information such a priori knowledge and temporal correlation can be utilized to improve sensing methods.


    Femtocell Clashes with Limited Number of Location Area Codes

    Femtocells are small cells of the future - they provide local coverage in areas where there is insufficient macro coverage and can increase overall network capacity. They connect to the cellular network via IP, and hence relies on a strong broadband infrastructure.


    Each femtocell requires a location area code (LAC) which essentially identifites itself to the macro network operator. Therefore, it is undesirable to have nearby or adjacent femtocells to have the same LAC. However, there are only a limited number of LACS to be given out. As such, there is a probability that in a given area, adjacent femtocells can have the same LAC. This probablity varies with inverse square of the number of LACs available.