MAC Layer Enhancements to Support IoT Traffic in Wireless Networks

Giluka, Mukesh Kumar and Tamma, Bheemarjuna Reddy (2018) MAC Layer Enhancements to Support IoT Traffic in Wireless Networks. PhD thesis, Indian Institute of Technology hyderabad.

Full text not available from this repository. (Request a copy)

Abstract

Today's world is of emerging Internet of Things (IoT) which has the capability to create a new trend of technological innovation by adding `smartness' to each component of various industries such as healthcare, transportation, home automation, and power grid systems. Presently, IoT is being supported by some existing communication technologies. But, its rapid growth will pose signi�cant challenges to the standardisation bodies to develop new technologies to support it e�ciently. Machine-to-machine communication (M2M) is one of the components of IoT which deals with networking and communication segment. The main characteristics of M2M commu- nication include large number of low-power, low-cost, low-range IoT/M2M devices, high uplink to downlink tra�c generation ratio, small packet sizes (e.g., 100 bytes), and infre- quent data generation. Traditional radio access technologies (RATs) such as LTE-A and Wi-Fi are mainly designed to support human-to-human communication (H2H) and human-to-machine com- munication (H2M). Examples of H2H/H2M applications include Internet browsing, smart- phone applications, VoIP, and video streaming. Because of several characteristic di�erences between M2M and H2H/H2M communication, incorporation of IoT/M2M poses many chal- lenges in di�erent layers of network protocol stack starting from application layer to physical layer. RATs mostly deal with physical layer (PHY) and medium access control layer (MAC), hence these two layers change with di�erent wireless technologies such as LTE-A, Wi-Fi, and Zigbee. This thesis focuses on MAC layer enhancements to support IoT/M2M tra�c in wireless networks. Medium access mechanisms of wireless networks can be broadly classi�ed into three ap- proaches viz., contention-free, contention-based, and hybrid. In contention-free approach, channel resources are exclusively allocated to devices to avoid collisions. TDMA, FDMA, CDMA and OFDMA are examples of the MAC layer protocols based on contention-free medium access mechanism. In contention-based approach, in order to occupy the chan- nel for data transmission, all devices need to contend for the channel which may result in collision when more than one device try to contend simultaneously. CSMA, ALOHA, Slotted ALOHA, and CSMA/CA are examples the MAC layer protocols which follow contention-based medium access mechanism. Hybrid approach inherits the concepts from both contention-based and contention-free approaches. Medium access mechanism followed in MAC layer of IEEE 802.11ah is an example of hybrid approach. In this thesis, we explore all three approaches and study the issues related to these approaches due to the incorpo- ration of IoT tra�c in contemporary networks. As a contribution, the thesis proposes solutions to some of the issues to facilitate IoT tra�c in these networks. Medium access mechanism in LTE-A networks follows contention-free approach. LTE-A networks were initially designed to support H2H/H2M tra�c but due to di�erent character- istics of IoT/M2M tra�c, its incorporation in these networks will impose several challenges. One of these challenges is e�cient radio resource allocation between H2H and M2M tra�c. As a solution, we propose three uplink packet scheduling schemes viz., class based priority (CBP) scheduling, class based dynamic priority (CBDP) scheduling, and enhanced class based dynamic priority scheduling (E-CBDP) schemes. These schemes classify both H2H and M2M tra�c among di�erent classes based on their real-timeness and reliability require- ments. Further, these classes are assigned priorities. Hence, priority of tra�c belonging to class-1 will be highest and it will decrease as the class number increases. CBP scheme separately allocates radio resources, also called as physical resource blocks (PRBs) to H2H and M2M tra�c. The scheme reserves a larger chunk of PRBs for H2H tra�c to ensure its quality of service (QoS). The main goal of the scheme is to maximise the QoS satisfaction of M2M tra�c without or least a�ecting the QoS of H2H tra�c. We evaluate the scheme classwise (both H2H and M2M tra�c are classi�ed into four classes) and compare its performance with some contemporary schemes based on the metrics delay, percentage of packet dropped, throughput, fairness, and energy consumption. The proposed scheme performs best for all the metrics of all classes of H2H and class-1 and class-2 of M2M but worst for class-3 and class-4 of M2M. In order to address the drawbacks of CBP scheme, the thesis proposes CBDP scheme which is di�erent from CBP scheme in the following ways: (i) no resource reservation for M2M and (ii) data packets are switched from low priority classes to high priority classes if not served within a particular time limit to avoid starvation. The simulation results show improvement in the performance of M2M. To enhance the performance of H2H, we further modi�ed CBDP scheme and proposed enhanced CBDP (E-CBDP) scheme. Here, switching of M2M tra�c to higher priority classes is dynamically controlled based on the performance of H2H tra�c in a particular time window. As a result, performance of H2H improves further. Downlink/uplink decoupling (DUDe) in LTE-A networks allows a device to perform uplink communication with one cell and downlink communication with another cell. In heterogeneous network scenarios where single Macro cell (eNodeB) and multiple small cells exist, DUDe can be used to facilitate M2M in the following two ways: (i) energy e�cient uplink communication where the M2M device connects in uplink with a nearer cell which is either Macro or small cell, thereby saves its energy and (ii) more D2D (device-to-device communication) pairs can be enabled under a cell due to shifting of some of the devices to another cell for uplink communication. In this thesis, we investigate the above advantages of DUDe favorable for M2M communication under contention-free approach of medium access. Our investigation reveals that spectral e�ciency increases and average transmit power per device decreases if DUDe is performed in the decoupling region (i.e., a region between Macro cell and small cell which is suitable for performing decoupling). Further, we observe that due to the reduction in the transmit power per device, interference in the vicinity of the devices decreases. As a result, more number of D2D pairs can be enabled which was not possible earlier due to higher interference created by the devices performing uplink with the Macro or small cell. Random access channel (RACH) procedure in LTE-A networks is a contention-based approach. As the large number of devices are associated with IoT/M2M communication, their simultaneous access to the network may cause RACH congestion which may lead to excessive delay in performing RACH procedure. Several solutions have been proposed in the literature to alleviate the RACH congestion caused by massive number of IoT/M2M devices. The main objective of these solutions is to provide uninterrupted connectivity to H2H or non-IoT devices as well as to save the network from reaching to a defunct state because of IoT/M2M devices. Extended access barring (EAB) is one among such solutions in which IoT/M2M devices are barred from accessing the network for an inde�nite amount of time when congestion in the network goes above a particular level. After studying di�erent EAB algorithms and analysing the performance of the standard EAB algorithm through extensive simulations, we reach to the conclusion that an ideal EAB based solution should have the following properties: (i) it should follow the EAB mechanism, (ii) impact of the solution on H2H devices should be analysed, and (iii) triggering condition for barring or unbarring of M2M devices should not be periodic but should be based on the tra�c load in the network. Most of the EAB based solutions in the literature lack some or all of above three properties. Apart from this, even if a solution has these properties, the algorithm to choose which class to bar or unbar signi�cantly a�ects the performance of EAB con�gured M2M devices. In this thesis, we propose an EAB based solution which has the above three properties. The proposed EAB algorithm implements a new method of barring or unbarring of M2M devices. By doing this, signifficant improvement can be achieved in the performance of M2M devices in terms of success rate, average end-to-end delay, and average number of backo�s. IEEE 802.11ah follows a hybrid approach of medium access to support M2M commu- nication in Wi-Fi networks. It is realized by providing contention-based access of channel to a particular set of IoT/M2M devices for a limited amount of time window, called as restricted access window (RAW). Further, in order to reduce the contention, the RAW is again divided into time slots called as RAW slots (of the order of tens of milliseconds) where a subset of IoT/M2M devices are allowed to perform contention-based access for their up- link data transmission. This thesis models the medium access mechanism of IEEE 802.11ah using a three dimensional Markov chain. The motivation for this modeling is because of the possibility that data transmission by a device may not finish within the RAW slot time. As a result, the device has to freeze its transmission and resume in next scheduled slot. The modeling covers this behavior of the MAC layer and provides a formula for normalized throughput. Further, simulation results show that throughput decreases with the increase in the number of IoT/M2M devices in a RAW slot. However, vary in RAW slot sizes has no signifficant impact on throughput.

[error in script]

IITH Creators: