IEEE Access (Jan 2017)

Packet-Size Optimization for Multiple-Input Multiple-Output Cognitive Radio Sensor Networks-Aided Internet of Things

  • Chitradeep Majumdar,
  • Doohwan Lee,
  • Aaqib Ashfaq Patel,
  • S. N. Merchant,
  • Uday B. Desai

DOI
https://doi.org/10.1109/ACCESS.2017.2687083
Journal volume & issue
Vol. 5
pp. 14419 – 14440

Abstract

Read online

The determination of optimal packet size (OPS) for a cognitive radio-assisted sensor networks (CRSNs) architecture is non-trivial. State of the art in this area describes various complex techniques to determine OPS for CRSNs. However, it is observed that under high interference from the surrounding users, it is not possible to determine a feasible OPS of data transmission under the simple point-to-point CRSN topology. This is contributed primarily to the peak transmit power constraint of the cognitive nodes. To address this specific challenge, this paper proposes a multiple-input multiple output-based CRSNs (MIMO-CRSNs) architecture for futuristic technologies, such as Internet of Things and machine-to-machine communications. A joint optimization problem is formulated, considering network constraints, such as the overall end-to-end latency, interference duration caused to the non-cognitive users, average BER, and transmit power. We propose our Algorithm 1 based on the generic exhaustive search technique to solve the optimization problem. Furthermore, a low complexity suboptimal Algorithm 2 based on solving classical Karush-Kuhn-Tucker conditions is proposed. These algorithms for MIMO-CRSNs are implemented in conjunction with two different channel access schemes. These channel access schemes are time-slotted distributed cognitive medium access control denoted as MIMO-DTS-CMAC and CSMA/CA-assisted centralized common control channel-based cognitive medium access control denoted as MIMO-CC-CMAC. Simulations reveal that the proposed MIMO-CRSN outperforms the conventional point-to-point CRSN in terms of overall energy consumption. Moreover, the proposed Algorithm 1 and Algorithm 2 show a perfect match, and the implementation complexity of Algorithm 2 is less than Algorithm 1. Algorithm 1 takes almost 680 ms to execute and provides OPS value for a given number of users, whereas Algorithm 2 takes 4-5 ms on average to find the OPS for the proposed MIMO-CRSN framework.

Keywords