<p>Wireless sensor networks are an advanced technology in which adaptive processing has received significant attention. A popular strategy in such adaptive networks is the incremental approach, as it necessitates less communication between nodes. However, in the presence of non-ideal noisy links, the performance of this strategy is severely degraded, since in each iteration, when the local estimates are sequentially circulated from one node to another, these exchanges are corrupted by the link noise. However, little effort has been made to improve the performance of incremental adaptive networks in such non-ideal conditions. This paper proposes an algorithm based on block processing methodology to improve the performance of these networks under noisy channel conditions. The exploitation of block processing is justified by reducing the overall number of exchanges between nodes, which in turn mitigates the effect of noisy links. Here, we will improve the performance of distributed block adaptive filtering in reducing the impact of noise by incorporating the concept of variable block length. The block length will be automatically adjusted during the adaptation process. Initially, this length is small, allowing for more communication between nodes and enabling the benefits of spatial diversity. As the adaptation process progresses, the block length gradually increases in inverse proportion to the norm of the block-error signal vector. This results in reduced exchange between nodes, leading to decreased link noise effects, and ultimately, more accurate steady-state estimation. The simulations demonstrate the superior performance of the proposed algorithm, especially under non-ideal link conditions.</p>