Jain, Abhishek (2014-08). Reliable Downlink Scheduling for Wireless Networks with Real-Time and Non-Real Time Clients. Master's Thesis. Thesis uri icon

abstract

  • In this thesis, we studied the problem of designing a down link scheduling policy to serve multiple types of clients from a base station in a time-varying wireless network. An ideal scheduling policy is fair among the clients, provides reliability to the clients, achieves high system throughput and prevents strategic clients from choosing incorrect means. The existing scheduling policies fail to achieve one or more of these features. The Proportional Fair scheduling policy for example, fails to provide reliability to the real time clients, while Round Robin policy provides reliability to the clients but fails to achieve high system throughput in a time-varying wireless network. Apart from these policies, there are scheduling policies which prioritize clients based on their delay requirements. Here, a client with lower priority may choose incorrect means like claiming false types of flows to obtain a better performance. A non-real time client may pretend to be a real time client if doing so, which might aid it to achieve better performance in terms of average throughput. We proposed a new scheduling policy that is not only proportionally fair but also provides reliability to the mixture of real time and non-real time clients over a shared wireless channel. Our proposed policy aims to serve clients with different service requirements and provides best service to the clients which furnish true information about their service requirements; the client claiming false service requirements is penalized with the reduced performance. We theoretically demonstrate the effectiveness of the algorithm by considering uniform distribution of service rates of all the clients. We then provide extensive simulation results of our scheduling policy under the fast fading Rayleigh model to show that this policy can be easily extended in wireless networks. We also show that our policy outperforms existing policies in providing better reliability to the clients and unlike other common policies, our policy degrades the performance of a client that chooses incorrect means.
  • In this thesis, we studied the problem of designing a down link scheduling policy to serve multiple types of clients from a base station in a time-varying wireless network. An ideal scheduling policy is fair among the clients, provides reliability to the clients, achieves high system throughput and prevents strategic clients from choosing incorrect means. The existing scheduling policies fail to achieve one or more of these features. The Proportional Fair scheduling policy for example, fails to provide reliability to the real time clients, while Round Robin policy provides reliability to the clients but fails to achieve high system throughput in a time-varying wireless network. Apart from these policies, there are scheduling policies which prioritize clients based on their delay requirements. Here, a client with lower priority may choose incorrect means like claiming false types of flows to obtain a better performance. A non-real time client may pretend to be a real time client if doing so, which might aid it to achieve better performance in terms of average throughput.

    We proposed a new scheduling policy that is not only proportionally fair but also provides reliability to the mixture of real time and non-real time clients over a shared wireless channel. Our proposed policy aims to serve clients with different service requirements and provides best service to the clients which furnish true information about their service requirements; the client claiming false service requirements is penalized with the reduced performance.

    We theoretically demonstrate the effectiveness of the algorithm by considering uniform distribution of service rates of all the clients. We then provide extensive simulation results of our scheduling policy under the fast fading Rayleigh model to show that this policy can be easily extended in wireless networks. We also show that our policy outperforms existing policies in providing better reliability to the clients and unlike other common policies, our policy degrades the performance of a client that chooses incorrect means.

publication date

  • August 2014