Amith V. Chincholi

Cross-Layer Optimization for Opportunistic Multi-MAC Aggregation

In this paper, we describe a RAT (radio access technology) agnostic approach referred to as OMMA (Opportunistic Multi-MAC Aggregation) that can aggregate multiple RATs operating over independent spectral bands. Aggregation and/or traffic shaping is performed above the air interface protocol stacks but below the IP layer. We investigate the problem of minimizing the average packet latency (the sum of queuing delay and serving delay) under certain constraints and propose an algorithm to compute optimal packet distribution ratio across RATs to be implemented by OMMA. Moreover, we statistically characterize the reordering delay for the OMMA system and propose an algorithm to compute a modified optimal packet distribution ratio which minimizes the packet end-to-end delay, i.e., the sum of average packet latency and average packet reordering delay. Finally, some numerical results are presented to study the average packet latency, average packet reordering delay and average packet end-to-end delay for different system parameters.

On the Packet Allocation of Multi-Band Aggregation Wireless Networks

Abstract The use of heterogeneous networks with multiple radio access technologies (RATs) is a system concept that both academia and industry are studying. In such system, integrated use of available multiple RATs is essential to achieve beyond additive throughput and connectivity gains using multi-dimensional diversity. This paper considers an aggregation module called opportunistic multi-MAC aggregation (OMMA). It resides between the IP layer and the air interface protocol stacks, common to all RATs in the device. We present a theoretical framework for such system while considering a special case of multi-RAT systems, i.e., a multi-band wireless LAN (WLAN) system. An optimal packet distribution approach is derived which minimizes the average packet latency (the sum of queueing delay and serving delay) over multiple bands. It supports multiple user terminals with different QoS classes simultaneously. We further propose a packet scheduling algorithm, OMMA Leaky Bucket, which minimizes the packet end-to-end delay, i.e., the sum of average packet latency and average packet reordering delay. We also describe the system architecture of the proposed OMMA system, which is applicable for the general case of the multi-RAT devices. It includes functional description, discovery and association processes, and dynamic RAT update management. We finally present simulation results for a multi-band WLAN system. It shows the performance gains of the proposed OMMA Leaky Bucket scheme in comparison to other existing packet scheduling mechanisms.