An adaptive hybrid schema for data-centric storage in wireless sensor networks

When should a storage strategy change?

When many events occur near a sensor network, a fixed storage layout can overload a few nodes. Adaptive resilient data-centric storage (AR-DCS) divides and merges storage zones as event frequency changes, balancing storage traffic while preserving replication.

Data-centric storage makes queries efficient, but fixed monitors and replicas can become hot spots. Their neighbors expend energy relaying traffic, eventually leaving stored information difficult to reach.

From the question to the evidence

1. Identify the storage hot spot

   Start with the data-centric storage trade-off. Mapping an event type to a known storage location makes queries easier, but heavy event traffic can exhaust the nodes that relay packets to that location. R-DCS adds monitors and replicas, yet a busy monitor can still become unreachable when its neighbors run out of energy.

   2016 AR-DCS article, PDF pp. 3-4, section 'AR-DCS'. Read source · PDF page 3 ↗

2. Estimate a threshold from the network

   Use sensing range and minimum node separation to estimate how many sensors can detect a target. Combine the average sensing count, node density, number of targets and current number of zones to obtain the proposed threshold. Equation (4) expresses this as average sensing count multiplied by density and target count, divided by zone count.

   2016 AR-DCS article, PDF pp. 4-5, threshold discussion, Table 1 and Eqs. (1)-(4). Read source · PDF page 4 ↗

3. Observe storage arrivals

   Each monitor samples the frequency of incoming storage packets. In the reported simulation, it evaluates whether a structural change is needed after every five received storage packets. This separates observation from reconfiguration and reduces repeated splitting and merging during rapid fluctuations.

   2016 AR-DCS article, PDF p. 4, 'Rapid swing in the frequency of occurrence of events'. Read source · PDF page 4 ↗

4. Split busy zones

   When the event frequency crosses the relevant threshold, monitors coordinate through the monitoring map. A parent zone divides into four subzones by halving its width and height. Each subzone gains a monitor and replicas, shortening the storage path and spreading traffic across more relay nodes.

   2016 AR-DCS article, PDF pp. 4 and 7, zone-division description and Figs. 1-4. Read source · PDF page 4 ↗

5. Merge zones as demand falls

   When storage arrivals return to a lower level, subzones merge into their parent. Fewer monitors and replicas then participate in querying. During aggregation, old subzone replicas retain their data for a transition period so that an in-flight query can still receive an answer. The article does not specify a universal duration for this period.

   2016 AR-DCS article, PDF p. 4, query handling during division and aggregation; PDF p. 7, zone aggregation. Read source · PDF page 4 ↗

6. Evaluate the storage-query trade-off

   Compare DCS, R-DCS and AR-DCS using the same target movements and scenario conditions. The study tracks storage packets, query packets, response packets and their total. Storage traffic improves when zones adapt, while querying more replicas adds traffic. The reported total is similar to basic DCS, with replication retained for recovery from node failure.

   2016 AR-DCS article, PDF pp. 6-9, simulation protocol and Figs. 4-7. Read source · PDF page 6 ↗

What the study reports

The discrete-event study uses a network of 150 stationary sensor nodes, with moving targets generating sensed events. It compares DCS, R-DCS, and AR-DCS under matched scenario conditions.

These are results reported by the source publication, not an independent reproduction. Conditions and source disagreements are retained beside the evidence they affect.

Scope & limitations

• Zone splitting adds query and response traffic and increases local processing complexity.
• The evaluation assumes stationary, similarly capable nodes and slowly changing radio conditions.
• Table and prose differ on simulation duration and initial energy; these values are not used as public headline facts.

Cited in context

Method outline

Explanatory pseudocode paraphrased from the published algorithm description. It is not executable research code. Threshold calculation, monitoring-map synchronization and data migration are represented at the level stated in the article.

INPUT network geometry, sensing range, node separation, targets and event stream
INITIALIZE the current zones, monitors and replicas using the data-centric layout
ESTIMATE the maximum sensors that can detect one target from the geometric model
DERIVE the average sensing count and node-density factor using Eqs. (1)-(3)
SET the adaptation threshold from sensing count, density, targets and zone count
FOR EACH storage packet arriving at a monitor:
    ROUTE the data through the current monitor and replica arrangement
    UPDATE the observed storage-arrival frequency
    AFTER every five received storage packets, EVALUATE zone adaptation
        IF frequency exceeds the relevant threshold:
            NOTIFY monitors through the monitoring map
            DIVIDE each affected parent layout into four subzones
            ASSIGN subzone monitors and replicas and transfer the stored data
        IF frequency returns to the lower operating level:
            MERGE subzones into their parent monitor and replica layout
            KEEP old subzone replicas available for the query transition period
FOR EACH query, USE the current mapping or a still-valid transition replica
TRACK storage, query, response and total packet counts to assess the trade-off

Cite this article

@article{Hejazi2016AdaptiveStorage,
  title = {An adaptive hybrid schema for data-centric storage in wireless sensor networks},
  author = {Hejazi, Pooya},
  journal = {International Journal of Distributed Sensor Networks},
  year = {2016},
  volume = {12},
  number = {10},
  doi = {10.1177/1550147716666665},
  url = {https://doi.org/10.1177/1550147716666665}
}

Download BibTeX · Download RIS

Article © its authors. CC BY 3.0. Original PDF and rights notices preserved.