A novel approach for energy- and memory-efficient data loss prevention to support Internet of Things networks

Where should a network keep a second copy?

A sensor network can lose information when a storage node runs out of battery or memory. Collaborative Memory and Energy Management (CoMEM) coordinates monitor and storage-node selection using both resources, combining data-centric retrieval with distributed storage.

Choosing storage nodes for energy alone can overlook memory exhaustion; choosing for memory alone can drain batteries. A practical data-dissemination method needs to account for both while balancing load.

From the question to the evidence

1. Treat memory and energy together

   A storage node may lose information because it runs out of memory, energy, or reachable neighbors. The paper models the WSN as a directed graph and combines memory use, energy consumption and a resource-dependent load-balancing factor in a cross-layer objective. It includes sensing, forwarding, querying, link capacity and flow-conservation constraints.

   2020 CoMEM article, PDF pp. 4-7, network model, Eqs. (1)-(22) and constraints C1-C17. Read source · PDF page 4 ↗

2. Establish zones and initial roles

   Partition the sensing region into zones and use geographic hashing to choose an initial monitor and storage nodes. The reported setup assumes stationary targets and a fixed zone count. Monitors provide known destinations for storage and queries, while storage nodes retain detailed event records.

   2020 CoMEM article, PDF pp. 7-8, DCS storage model and the CoMEM initialization description. Read source · PDF page 7 ↗

3. Share resource information

   Nodes report normalized memory and energy information to their zone monitor, either along with sensed data or through dedicated packets. The monitor maintains a performance table and computes an election metric using both resources. A variation test triggers re-election instead of restricting replica selection to the current storage node's immediate neighbors.

   2020 CoMEM article, PDF p. 8, performance table and Eqs. (23)-(25). Read source · PDF page 8 ↗

4. Re-elect storage nodes across the zone

   The election procedure compares candidates using energy and memory coefficients. It builds coverage counts, excludes candidates covered by a stronger candidate in both dimensions, and selects a candidate with the highest coverage count. Figure 4 resolves ties by the energy coefficient and then the memory coefficient. This is a heuristic selection rule, not a proof of global optimality.

   2020 CoMEM article, PDF pp. 8-9, replica election discussion and Fig. 4. Read source · PDF page 8 ↗

5. Rotate the monitor when needed

   Storage nodes receive the zone's residual-energy information and evaluate the monitor's energy gap. When the paper's threshold test is met, the storage node announces a new monitor with the highest reported residual energy. The outgoing monitor transfers its records to the replacement. This addresses the resource cost of concentrating coordination traffic at one monitor.

   2020 CoMEM article, PDF p. 9, monitor-election mechanism and Eq. (26). Read source · PDF page 9 ↗

6. Compare the full trade-off

   Evaluate CoMEM against DCS, R-DCS and LCDDR using energy consumption, the combined objective, delay, data loss, computational work and scalability. The optimization problem is solved with a genetic algorithm to provide the study's optimization-based comparison. CoMEM leads on several resource measures, while basic DCS has lower delay and less computational work in the reported ranking.

   2020 CoMEM article, PDF pp. 11-15, performance evaluation, Figs. 6-11, Table 3 and conclusions. Read source · PDF page 11 ↗

What the study reports

The principal evaluation models static nodes in an IEEE 802.15.4/6LoWPAN network, with 30 nodes in a 12 m by 12 m area and one to ten targets. Mobility and other wireless technologies are outside the evaluated scope.

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

• The evaluation assumes static nodes and relatively stable radio conditions.
• Extra coordination raises delay compared with basic data-centric storage.
• The reported comparisons depend on the study traffic, energy, memory, and physical-layer models; no universal percentage improvement is claimed.

Cited in context

Method outline

Conceptual pseudocode paraphrased from the CoMEM description and Figure 4. It is not the published implementation or a directly executable algorithm. Resource normalization and threshold tests remain as defined in the paper; the source's resource-coefficient notation requires care when implementing.

INPUT zone membership, event records and node memory/energy reports
INITIALIZE a monitor and storage nodes in each zone using geographic hashing
ON sensed data, SEND the record to the zone monitor
    KEEP the summary at the monitor and FORWARD details to the assigned storage node
UPDATE the monitor's table with normalized resource reports from zone members
COMPUTE the joint election metric and its variation test from Eqs. (23)-(25)
WHEN the storage re-election condition holds:
    INITIALIZE each candidate's coverage count to one
    COMPARE each candidate with active candidates in its zone
    IF another candidate has larger energy and memory coefficients:
        MARK the weaker candidate covered and INCREMENT the stronger candidate's count
    SELECT the highest coverage count, then energy coefficient, then memory coefficient
    USE the elected candidate for the storage/replica role
PERIODICALLY SHARE residual-energy information with the storage node
WHEN the monitor energy-gap test in Eq. (26) exceeds its threshold:
    ANNOUNCE the highest-residual-energy candidate as the new monitor
    TRANSFER the outgoing monitor's records to the new monitor
ON a query, LET the monitor answer from its summary or contact the relevant storage node
EVALUATE resource use, delay and data loss together across the study scenarios

Cite this article

@article{Hejazi2020IotDataLoss,
  title = {A novel approach for energy- and memory-efficient data loss prevention to support Internet of Things networks},
  author = {Hejazi, Pooya and Ferrari, Gianluigi},
  journal = {International Journal of Distributed Sensor Networks},
  year = {2020},
  volume = {16},
  number = {6},
  doi = {10.1177/1550147720929823},
  url = {https://doi.org/10.1177/1550147720929823}
}

Download BibTeX · Download RIS

Article © its authors. CC BY 4.0. Original PDF and rights notices preserved.