Wireless Sensor Networks (WSNs) have gained popularity in a lot of emerging areas since their evolution, like the monitoring of natural phenomena concerning geosciences in various disciplines. Some relevant areas include the monitoring of oil and gas exploration field, earth quake and volcanic eruptions etc. In these processes real time phenomena are monitored in remote fields and after data collection through WSN, it is ported to the far flung research centers for further investigation and decision making. Here 2-tier WSN is considered with lower tier of sensor nodes in WiFi and upper tier of WiMAX as a backhaul for data transportation to distant research facility. However WiMAX has already popularity for applications with main configuration of downlink data delivery direction, like serving internet hot spots and similar amenities. This innovative research work is concerned with previously mentioned phenomena monitoring which needs high data throughput efficiency in the uplink direction. For dense sensor node concentration, in order to transport consolidated output form all sensor nodes in real time phenomena of impulsive nature, sufficient uplink throughput is needed with low latency which forms a bottleneck in these cases for WiMAX backhaul. Real time applications constrain end-to-end delay and hence throughput which severely affect the performance and the accuracy of the monitoring. To the system only fixed bandwidth is available, for which resources of slots have to be shared between downlink and uplink. Here WiMAX OFDM in TDD mode is considered. Initially analysis has been presented to point out main important constituent parameters of WiMAX which contribute to throughput. This research work proposes a solution to enhance uplink bandwidth allocation efficiency for these phenomena through adaptive shift of WiMAX frame ratio. In this regard for WiMAX adaptive ratio shift researchers have used a number of scenarios but all of them are not very efficient and have drawbacks. Some of them are using either fixed margins or some other are using fixed step sizes for upward or downward (increment) without any calculation. If a link (uplink or downlink) is running near full capacity then any kind of sudden appearance of internal system overheads in the form of system message or broadcast may bring the link to fragmentation. And fragmentation successively may contribute to increase the link overheads further, causing either increased delay or some other problem related to packet drop, re-transmission or transmission failure. If problems are induced, this may take relatively longer at a reduced data transfer rate. In order to avoid similar problems, relevant calculations have to be performed to guarantee good QoS values of maximum link throughput while keeping low delay and packet drop, which needs adequately more resources. In fact keeping in view the complex trade-off between QoS parameters and system resources, the optimization problem is formulized to maximize the uplink throughput while keeping the latency and packet drop of downlink to minimum limit, to facilitate the efficient operation of uplink’s momentarily bulky traffic. A novel solution to this problem is obtained through the incorporation of stochastic processes with random variable in finite state space. The analytical and mathematical expressions are contributed for the different analytical models. A MMPP traffic model is formed for OFDMA transmission. This is supported with a DTMC system model for queuing theoretic performance modeling. Analytical and numerical values of performance parameters like throughput, packet delay and probability of packet drop are estimated for resource allocation through mathematical models of stochastic process. First of all to restrict the downlink traffic to minimum level a traffic pattern is defined from downlink to WSN sensor nodes. While considering packet scheduling, two scenarios are taken into account, one is concerned with known MCS distribution and the other is concerned with unknown MCS distribution. In this process mainly affecting factors of frame overheads are also taken into account. Then by using all these analytic and mathematical models, an algorithm is formulated to find out minimum optimal resource requirement on down link after considering these QoS parameters. Through this manipulation rest of the resources can be transferred to facilitate uplink operation. Another DTMC model is designed to check and quantify the uplink frame utilization. When uplink frame utilization is beyond normal, the frame resources have to be incremented to previously calculated maximum limit by adaptively remapping the frame ratio. Also when utilization on uplink reduces below normal, it is remapped to normal frame ratio. The main algorithm is formed by appropriately including all these previous analytical models. This sets the adaptive ratio to the best suitable maximum value, to facilitate the uplink bulky traffic, and saves the link from congestion and slowing down. This also ensures minimization of previously mentioned errors of delay, drop or timeout related problems. This process achieves highest degree of convergence in just one step by providing maximum throughput on uplink without degrading QoS parameters on the downlink operation. In order to prove the results firstly analytical results are obtained from the computations of the algorithm in MATLAB. Secondly intensive simulations are conducted around Rayleigh flat fading channel through models in three steps. One simulation model gives results for downlink minimum traffic, next model is for uplink maximum traffic and another model is for uplink normal ratio traffic. All these results together prove the accuracy and superiority of the algorithm by showing an outstanding uplink bandwidth efficiency enhancement without degrading the downlink operation. Future research directions are to further enhance the analytical models for more states. Also more variations to overhead analysis can be added by more realistic models and with more overhead reduction techniques.