Energy Protocol (EP) Overview

• Energy Protocol (EP) is a family of decentralized frameworks enabling energy exchange through networking-style abstractions and protocol layers.
• It integrates software-defined control planes, secure routing, and real-time flow management across DC microgrids and energy networks.
• EP also supports blockchain-based and standardization approaches to facilitate privacy-preserving, location-aware peer-to-peer trading and settlement.

Energy Protocol (EP) is a term used in recent energy-systems literature for protocol frameworks that coordinate decentralized energy exchange through networking-style abstractions. In "EnergyNet Explained: Internetification of Energy Distribution," EP is the open protocol implemented by EP-Server instances to mediate software-controlled energy flows across Energy Local Area Networks (ELANs) and Energy Wide Area Networks (EWANs) built on DC microgrids (Birgersson et al., 9 Sep 2025). In "Lightweight Blockchain Framework for Location-aware Peer-to-Peer Energy Trading," EP denotes a lightweight, location-aware, privacy-preserving, fully decentralized peer-to-peer energy trading protocol spanning off-chain advertisement, on-chain negotiation and settlement, reputation, and Anonymous Proof of Location (Khorasany et al., 2020). In "Energy Internet: A Standardization-Based Blueprint Design," related protocol functionality is expressed through standardized Block of Energy Exchange (BEE) transfer over Energy IP, Energy TCP, and an Energy HTTP-like application layer, mediated by the Energy Internet Card (Guo et al., 2024). The literature therefore uses EP to describe interoperable control, trading, and settlement mechanisms that align energy distribution with communication-network design.

1. Scope and terminological usage

The term EP does not refer to a single universally standardized wire format across the cited literature. Instead, it names distinct but structurally related protocol designs for decentralized energy coordination. One lineage emphasizes control-plane routing over DC microgrids; another emphasizes privacy-preserving market settlement on blockchain infrastructure; a third emphasizes standardization of energy-transfer records and transport semantics.

Work	EP object	Core artifacts
(Birgersson et al., 9 Sep 2025)	Open control protocol in EnergyNet	Energy Router, EROS, EP-Server, ENMS, ELAN/EWAN
(Khorasany et al., 2020)	Location-aware, privacy-preserving P2P trading protocol	AD, EN/LP/EI transactions, DR contract, A-PoL
(Guo et al., 2024)	Protocols supporting BEE transfer	BEE, Energy IP, Energy TCP, Energy HTTP-like, EIC

This variation in meaning is important for technical interpretation. In one formulation, EP primarily allocates and routes physical energy flows between routers; in another, it clears bilateral trades while protecting identity and location; in the third, it standardizes the data structures and transport procedures that accompany energy delivery. A plausible implication is that "EP" in current research is better read as an architectural family of energy-network protocols than as a single settled standard.

2. EnergyNet EP as a software-defined control plane

In EnergyNet, EP sits inside a software-defined control plane superimposed on a DC microgrid fabric. Each Energy Router contains two redundant controller modules: EROS (Energy Router Operating System), responsible for local port-to-port routing logic, and EP-Server, responsible for inter-router negotiation. EP messages flow over mutually authenticated, TLS-protected channels between EP-Server instances. At the hardware and interface level, router ports are organized as A-consumption, B-legacy grid, C-peer routers, and D-local resources, all interconnected by an internal DC backplane under EROS control. EP-Server mediates port-to-port flows across ELAN and EWAN boundaries, while ENMS (Energy Network Management System) orchestrates large fleets of EP-enabled routers, pushes policies in a BSS/OSS/eTOM style, and monitors health at operator scale (Birgersson et al., 9 Sep 2025).

The same work explicitly separates architectural contribution from contingent outcomes. Its Tier-1 contribution consists of components, interfaces, and operating model. Tier-2 contains expected outcomes contingent on adoption, including local-first autonomy with global interoperability, near-real-time operation with local buffering, removal of EV-charging bottlenecks, freed grid capacity for data centers and industrial electrification, and a trend toward low, predictable, fixed-cost clean energy. This distinction matters because it separates what the protocol specifies from what large-scale deployment may later deliver.

Galvanic separation is central to the EnergyNet formulation. The Energy Router enforces galvanic separation and uses software-controlled energy flows over a DC backplane. The paper’s narrative synthesis further characterizes galvanic separation as a "firewall" for frequency stability and presents local buffering as a replacement for brittle real-time strictures. In this interpretation, EP is not only a signaling layer; it is tightly coupled to converter-mediated physical routing and to the operating constraints of decentralized power electronics.

3. EnergyNet signaling, state machines, and optimization model

EnergyNet summarizes a canonical EP packet structure through an EP_Header with Version:8, Flags:8, MsgType:8, Reserved:8, Length:16, MsgID:32, SrcID:64, and DstID:64. The defined message types are HELLO, LINK_STATE, ROUTE_REQ, ROUTE_REP, FLOW_REQ, FLOW_REP, POLICY_UPD, HEARTBEAT, and ERROR. FLOW_REQ carries fields such as ReqPower, StartTime, Duration, Priority, PolicyTag, and a variable-length digital signature. The protocol version is currently 0x01, and SrcID/DstID are globally unique router identifiers, for example IEEE EUI-64 (Birgersson et al., 9 Sep 2025).

Three EP procedures are identified: neighbor discovery and link-state exchange, route discovery and path selection, and flow admission control. The pseudocode sketch defines states DISCOVERING, IDLE, WAIT_ROUTE_REP, WAIT_FLOW_REP, and ACTIVE_FLOW. A router starts by sending HELLO on all known links, learns neighbors through HELLO and LINK_STATE, issues ROUTE_REQ{DstID, MinCap, Priority} upon an application REQUEST_FLOW, evaluates ROUTE_REP{Path, TotalCost} against a policy_limit, and, on successful FLOW_REP{Accept}, configures the EROS backplane to pin the route. Flow teardown returns the state machine to IDLE. The paper states that these procedures can be represented as finite-state machines with transitions triggered by reception of control messages, timeouts, and errors.

For EWAN routing, EnergyNet adopts a multicommodity flow model on a logical router graph $G=(V,E)$ with link capacities $C_{ij}$ and unit costs $c_{ij}$:

$$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$

Here $s_i>0$ is net injection at router $i$, defined as local generation minus consumption. In practice, the protocol extends this model with priority weights $w_i$ and dynamic costs $c_{ij}(t)$ derived from price or network-health signals.

Security, reliability, and QoS are specified at the control-plane level. EP uses Mutual-TLS authentication for all channels, with X.509 certificates provisioned by ENMS; message confidentiality and integrity are provided by TLS together with ECDSA signatures on critical control messages; Role-Based Access Control tags ports for A/B/C/D side logic; critical messages carry MsgID and AckReq flags and are retransmitted up to $N=3$ times if unanswered; and the Flags field reserves 3 bits for priority, with FIFO or strict-priority scheduling for FLOW_REP acknowledgments and backplane grants. The paper’s narrative description also associates EROS with sub-second port switching and EP-Server with policy-driven inter-router negotiation, reinforcing the view that EP is an operational real-time control protocol rather than only a market API.

4. Blockchain-based EP for decentralized peer-to-peer trading

In the blockchain-based formulation, EP is a location-aware, privacy-preserving, fully decentralized peer-to-peer energy trading protocol. The physical layer consists of a radial or meshed distribution network with lines $\mathcal{L}$, a trusted grid operator that maintains the real topology, issues grid prices $C_{ij}$0, computes service charges, and hosts an off-chain Advertisement Database (AD). The overlay layer contains producer agents $C_{ij}$1 and consumer agents $C_{ij}$2, each with a tamper-resistant smart meter and a public-private key pair, plus a public blockchain that stores finalized Energy Negotiation (EN), Late-Payment (LP), and Energy-Injection (EI) transactions and a Dispute-Resolution (DR) smart contract (Khorasany et al., 2020).

The market-settlement workflow has five stated stages. First, agents compute surplus or deficit $C_{ij}$3 using the smart meter. Second, producers and consumers post Advertisement Transactions to AD:

$C_{ij}$4

Third, each agent reads AD, computes a priority index $C_{ij}$5, and partitions potential partners into groups $C_{ij}$6. Fourth, consumers initiate EN transactions over the blockchain overlay with the highest-priority group, and producers and consumers iteratively solve local optimization subproblems while exchanging only $C_{ij}$7 and $C_{ij}$8; when both agreement flags are 1, the last EN is recorded on-chain. Fifth, the consumer issues LP, the producer injects energy and issues EI, and atomic commit rules link LP and EI so that both either commit or abort.

The global objective is a social-welfare maximization with grid service charges:

$C_{ij}$9

subject to producer and consumer energy bounds and the equality constraint $c_{ij}$0. The solution method is dual decomposition with subgradient-projection. At each iteration, each producer updates $c_{ij}$1 and $c_{ij}$2, while consumers perform symmetric updates; critically, only $c_{ij}$3 and $c_{ij}$4 are exchanged peer-to-peer, whereas cost and utility parameters remain local.

Pricing is explicitly location-aware. The per-unit grid service charge is

$c_{ij}$5

where $c_{ij}$6 is a fixed fee in ¢/kWh/km and the electrical distance is

$c_{ij}$7

with $c_{ij}$8 the PTDF of line $c_{ij}$9 for a transaction $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$0. Price bounds satisfy

$$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$1

Agents also maintain a reputation score $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$2, and if delivered energy is less than committed energy, the DR contract computes the delivery ratio $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$3 and updates reputation according to

$$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$4

Trading-partner selection combines proximity and reputation through

$$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$5

Agents sort by $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$6, partition into $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$7 groups, and negotiate group by group. The protocol’s privacy layer is the Anonymous Proof of Location (A-PoL), which uses standard public-key signatures and Merkle-tree membership proofs. A smart meter receives a key pair from the grid operator, forms a Merkle root over one-time public keys, requests a Certificate of Location, and later presents an anonymous location credential

$$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$8

Verification checks Merkle membership, one-time-key signature validity, the verifier’s signature on $$\min_{\{f_{ij}\}} \sum_{(i,j)\in E} c_{ij} f_{ij} \quad \text{s.t.} \quad \sum_j f_{ij} - \sum_k f_{ki} = s_i,\ \forall i\in V, \qquad 0 \le f_{ij} \le C_{ij},\ \forall (i,j)\in E.$$9, and the verifier key’s authenticity. The paper’s security analysis attributes privacy to verifier randomization and one-time-key unlinkability, prevents location forging through genuine smart-meter verification, and prevents double-spending through the one-to-one EN/LP/EI reference structure and atomic commit rules.

5. Standardization-based EP through BEE transfer and the Energy Internet Card

The standardization-based blueprint organizes energy exchange around the Block of Energy Exchange (BEE), the fundamental unit of exchange in the Energy Internet. A BEE has two stages: an ex-ante data stage, in which the terms of a prospective energy exchange are committed to a standard nine-field record, and a real-time delivery stage, in which pledged energy and any associated certificates or carbon rights are physically moved. The nine data-header entries are BEE_ID, Src_MAC, Dst_MAC, Src_IP, Dst_IP, Time_Interval, Energy_Quantity, Price, and Emission_Profile. Formally,

$s_i>0$0

with total length bounded by the TCP maximum segment size or by the IP fragmentation limit of approximately $s_i>0$1 bytes (Guo et al., 2024).

The protocol stack follows a simplified four-layer TCP/IP model with two major modifications. At the link layer, physical connections carry both the information header and the actual energy. At the network layer, Energy IP assigns an address to each node and performs forwarding, routing, and static flow control through per-source or per-destination quantity caps in the IP header’s options field. At the transport layer, Energy TCP mirrors classic TCP through a three-way handshake:

• Client to server: SYN (seq=x, quantity_cap=Q₁)
• Server to client: SYN-ACK (seq=y, ack=x+1, quantity_cap=Q₂)
• Client to server: ACK (ack=y+1)

Flow control is enforced by a sliding window expressed in kWh rather than bytes; the receiver acknowledges cumulative energy delivered so far. Error handling also parallels standard TCP: timeouts cause retransmission of unacknowledged BEE headers, and corrupted headers that fail CRC are re-requested. At the application layer, BEEs are encapsulated in an HTTP-style POST or PUT whose body contains exactly nine fields in a JSON-style array, and successful receipt yields 200 OK before physical delivery begins. The Energy TCP connection state machine is identical to classic TCP’s, including the ESTABLISHED state and eventual closure through a FIN sequence.

The Energy Internet Card (EIC) is the mandatory hardware/software module for each participant. Its functions are to encapsulate identity in a fixed 48-bit MAC-like address, parse incoming BEE headers, enforce static and dynamic delivery limits, forward or block physical energy flows accordingly, and automatically update the local client profile—energy inventory, certificates, carbon rights, and trading record—once a BEE completes. Each EIC has a permanent 48-bit EIC-MAC and may acquire a dynamic Energy IP address when joining a Virtual Power Plant or moving to a different microgrid. The profile-update process involves parsing the nine fields, validating digital signatures, enforcing per-interval quantity caps in the TCP window option, opening real-time energy coupling, and appending the completed trade to a local profile ledger; if configured, the record can be propagated to a distributed-ledger or smart-contract platform.

At the application layer, each VPP or aggregator hosts a BEE Pool containing outstanding Buy and Sell offers, each carrying the nine-field BEE header plus a timestamp. A simple stack-based matching engine finds compatible offers by time slot and price, merges them into an agreed BEE, initiates the Energy TCP handshake, removes matched offers on success, and reinserts unmatched remainders on failure. After settlement, the Energy ISP observes only aggregated net nodal injections,

$s_i>0$2

and uses these quantities in ACOPF to dispatch centralized resources. The blueprint therefore explicitly decouples user-side trading from centralized dispatch: participants negotiate P2P trades among themselves, while the Energy ISP enforces static and dynamic quantity caps and dispatches central resources without needing to know bilateral trading details.

6. Demonstrators, performance, limitations, and technical interpretation

The cited literature reports both demonstrators and simulation-based results. In the Tamarinden demonstrator in Örebro and the "SWS EnergyNet-0" pilot in Lund, EP handshake latency from HELLO to LINK_STATE was approximately $s_i>0$3 ms in a single ELAN, route discovery from ROUTE_REQ to ROUTE_REP over EWAN was less than $s_i>0$4 ms, and flow-admission round-trip time from FLOW_REQ to FLOW_REP was less than $s_i>0$5 ms on 800\,\mathrm{VDC} Freedom Cable. Under upstream grid blackout, local EP controllers maintained prioritized "some power" for more than $s_i>0$6 h in lab tests. Tamarinden reportedly achieved a $s_i>0$7 reduction in peak grid draw and $s_i>0$8 lower annual peak power, attributed directly to EP-coordinated microgrid routing and local markets. The same narrative states that robust mTLS-backed messaging and sub-$s_i>0$9 ms control-plane interactions allow a neighborhood of $i$0 homes to behave as a single, software-defined power fabric (Birgersson et al., 9 Sep 2025).

The standardization-based blueprint reports a four-node radial-network test case over a 24 h horizon with 12 two-hour periods. Compared with traditional dispatch, the Energy Internet formulation yielded social welfare up by $i$1, system carbon emissions down by $i$2 with wind and solar curtailment eliminated, and grid surplus down by $i$3 while remaining nonnegative and meeting the $i$4 margin target. Renewable, battery, and EV profits increased dramatically, consumer surplus increased in both urban and rural areas, and losses increased slightly because individual resources did not know the full network topology. The paper attributes these results to autonomous resource adjustments following real-time prices discovered in the P2P market and frames the distributional change as a transfer of surplus from centralized operators to end users (Guo et al., 2024).

These results also delimit common misconceptions. First, EP is not merely a market-clearing message format. In EnergyNet it extends into router-state machines, QoS, admission control, and EROS backplane configuration. In the blockchain framework it includes settlement atomicity, dispute resolution, reputation, and privacy-preserving location proofs. In the standardization blueprint it includes transport semantics, device identity, flow-control caps, and profile updates. Second, the literature distinguishes architectural specification from contingent socio-technical outcomes. EnergyNet explicitly marks Tier-2 outcomes such as removal of EV-charging bottlenecks or low, predictable, fixed-cost clean energy as dependent on adoption rather than as intrinsic properties of the Tier-1 protocol. Third, "packetized energy" is used analogically rather than as a claim that electrical power is discretized in the same way as digital packets: the EnergyNet narrative instead states that flows have bandwidth in power and duration in time windows, and that software-defined routing is combined with local buffering and galvanic separation.

Across these formulations, EP consistently imports networking primitives—addressing, discovery, link-state exchange, flow control, authentication, retransmission, handshake logic, and layered abstraction—into decentralized energy systems. The principal technical divergence lies in what is being protocolized: control-plane routing of physical flows, peer-to-peer market settlement under privacy constraints, or standardized exchange blocks carried over an energy-specific transport stack. The literature collectively suggests that the central research problem is not only energy scheduling, but the co-design of communication semantics, device identities, control authority, and settlement logic for systems in which distributed resources, local autonomy, and operator-scale coordination must coexist.